Merge branch 'master' into ep/disable-background

* smp server: log prometheus metrics

* save metrics

* diff

* lines

* version

* do not include Prometheus into client

* corrections

Co-authored-by: sh <37271604+shumvgolove@users.noreply.github.com>

* corrections

Co-authored-by: sh <37271604+shumvgolove@users.noreply.github.com>

* corrections

Co-authored-by: sh <37271604+shumvgolove@users.noreply.github.com>

* add timestamp to metrics

* remove type

* remove version

---------

Co-authored-by: sh <37271604+shumvgolove@users.noreply.github.com>

.github/workflows/build.yml

@@ -63,7 +63,7 @@ jobs:
           mv $(cabal list-bin xftp) xftp-ubuntu-${{ matrix.platform_name}}
 
       - name: Build changelog
-        if: startsWith(github.ref, 'refs/tags/v') && matrix.os == 'ubuntu-22.04'
+        if: startsWith(github.ref, 'refs/tags/v')
         id: build_changelog
         uses: mikepenz/release-changelog-builder-action@v1
         with:

CHANGELOG.md

@@ -1,3 +1,156 @@
+# 6.2.0
+
+Version 6.2.0.7
+
+Build:
+- client_library flag to build only used modules in the clients, remove package yaml
+
+SMP server:
+- journal storage for messages (BETA).
+- prevent race condition when deleting queue and to avoid "orphan" messages (#1395).
+
+SMP agent:
+- support SMP and XFTP server roles (storage/proxy) and operators (#1343).
+- treat blocked STM and other critical errors that offer restart as temporary for message delivery (#1405).
+- fix inconsistent state after app restart while accepting contact request (#1412).
+
+# 6.1.3
+
+SMP server: fix restoring notification credentials.
+
+# 6.1.2
+
+Servers: more reliable restoring of state.
+
+SMP server: reduced memory usage and faster start.
+
+Notifications: compensate for iOS notifications being droppted by Apple while device is offline (#1378):
+- Ntf server: send multiple SMP notifications in one iOS notification.
+- Agent: get multiple messages for one iOS notification.
+
+# 6.1.1
+
+SMP:
+- stop server faster (#1371)
+- add STORE error (#1372)
+
+# 6.1.0
+
+Version 6.1.0.7
+
+SMP server and client:
+- transport block encryption (#1317).
+
+Agent:
+- batch and optimize iOS notifications processing (#1308, #1311, #1313, #1316, #1330, #1331, #1333, #1337, #1346).
+- allow receiving multiple messages from single iOS notification (#1355, #1362).
+- prepare connection to accept to avoid race condition with events (#1365).
+- transport isolation mode "Session" (default) to use new SOCKS credentials when client restarts or SOCKS proxy configuration changes (#1321).
+
+Ntf server:
+- control port (#1354).
+- enable pings on ntf subscriptions, to resubscribe on reconnection (#1353).
+
+SMP server:
+- support multiple server ports (#1319).
+- support serving HTTPS and SMP transport on the same port (#1326, #1327).
+- persist iOS notifications to avoid losing them when Ntf server is offline (#1336, #1339, #1350).
+- fix lost notification subscriptions (#1347).
+- reject SKEY with different key earlier, at verification step (#1366).
+- pass server information via CLI during server initialization (#1356).
+- show version on server page (#1341).
+- explicit graceful shutdown on SIGINT (#1360).
+
+XRCP (remote access protocol):
+- use SHA3-256 in hybrid key agreement (#1302).
+- session encryption with forward secrecy (#1328).
+
+# 6.0.5
+
+SMP agent:
+- support generic SOCKS proxy (without isolate-by-auth).
+- reduce max message sizes
+
+# 6.0.4
+
+SMP server:
+- better performance/memory: fewer map updates on re-subscriptions (#1297), split and reduce STM transactions (#1294)
+- send DELD when subscribed queue is deleted (#1312)
+- add created/updated/used date to queues to manage expiration (#1306)
+
+XFTP server: truncate file creation time to 1 hour (#1310)
+
+Servers:
+- bind control port only to 127.0.0.1 for better security in case of firewall misconfiguration (#1280)
+- reduce memory used for period stats (#1298)
+
+Agent: process last notification from list (#1307)
+- report receive file error with redirected file ID, when redirect is present (#1304)
+- special error when deleted user record is not in database (#1303)
+- fix race when sending a message to the deleted connection (#1296)
+- support for multiple messages in a single notification
+
+Ntf server:
+- only use SOCKS proxy for servers without public address (#1314)
+
+# 6.0.3
+
+Agent:
+- fix possible stuck queue rotation (#1290).
+
+SMP server:
+- batch END responses when subscribed client switches to reduce server and client traffic.
+- reduce STM transactions for better performance.
+- add stats for END events and for SUB/DEL event batches.
+- remove "expensive" stats to save memory.
+
+# 6.0.2
+
+SMP agent:
+- fix stuck connection commands when a server is not responding.
+- store query errors, reduce slow query threshold to 1ms.
+
+Notification server:
+- reduce PING interval to 1 minute.
+- fix subscriptions disabled on race condition (only mark subscriptions with END status when received via the active connection).
+
+# 6.0.1
+
+SMP agent:
+- support changing user of the new connection.
+- do not start delivery workers when there are no messages to deliver.
+- enable notifications for all connections.
+- combine database transactions when subscribing.
+
+SMP server:
+- safe compacting of store log.
+- fix possible race when creating client that might lead to memory leak.
+
+Dependencies: upgrade tls to 1.9
+
+# 6.0.0
+
+Version 6.0.0.8
+
+Agent:
+- enabled fast handshake support.
+- batch-send multiple messages in each connection.
+- resume subscriptions as soon as agent moves to foreground or as network connection resumes.
+- "known" servers to determine whether to use SMP proxy.
+- retry on SMP proxy NO_SESSION error.
+- fixes to notification subscriptions.
+- persistent server statistics.
+- better concurrency.
+
+SMP server:
+- reduce threads usage.
+- additional statistics.
+- improve disabling inactive clients.
+- additional control port commands for monitoring.
+
+Notification server:
+- support onion-only SMP servers.
+
 # 5.8.2
 
 Agent:

README.md

@@ -149,8 +149,15 @@ On Linux, you can deploy smp and xftp server using Docker. This will download im
 You can install and setup servers automatically using our script:
 
 ```sh
-curl --proto '=https' --tlsv1.2 -sSf https://raw.githubusercontent.com/simplex-chat/simplexmq/stable/install.sh -o simplex-server-install.sh \
-&& if echo 'b8cf2be103f21f9461d9a500bcd3db06ab7d01d68871b07f4bd245195cbead1d simplex-server-install.sh' | sha256sum -c; then chmod +x ./simplex-server-install.sh && ./simplex-server-install.sh; rm ./simplex-server-install.sh; else echo "SHA-256 checksum is incorrect!" && rm ./simplex-server-install.sh; fi
+curl --proto '=https' --tlsv1.2 -sSf https://raw.githubusercontent.com/simplex-chat/simplexmq/stable/install.sh -o simplex-server-install.sh &&\
+if echo '53fcdb4ceab324316e2c4cda7e84dbbb344f32550a65975a7895425e5a1be757 simplex-server-install.sh' | sha256sum -c; then
+  chmod +x ./simplex-server-install.sh
+  ./simplex-server-install.sh
+  rm ./simplex-server-install.sh
+else
+  echo "SHA-256 checksum is incorrect!"
+  rm ./simplex-server-install.sh
+fi
 ```
 
 ### Build from source

apps/ntf-server/Main.hs

@@ -15,7 +15,7 @@ logCfg = LogConfig {lc_file = Nothing, lc_stderr = True}
 
 main :: IO ()
 main = do
-  setLogLevel LogDebug -- change to LogError in production
+  setLogLevel LogInfo
   cfgPath <- getEnvPath "NTF_SERVER_CFG_PATH" defaultCfgPath
   logPath <- getEnvPath "NTF_SERVER_LOG_PATH" defaultLogPath
   withGlobalLogging logCfg $ ntfServerCLI cfgPath logPath

apps/smp-server/Main.hs

@@ -19,4 +19,4 @@ main = do
   setLogLevel LogDebug
   cfgPath <- getEnvPath "SMP_SERVER_CFG_PATH" defaultCfgPath
   logPath <- getEnvPath "SMP_SERVER_LOG_PATH" defaultLogPath
-  withGlobalLogging logCfg $ smpServerCLI_ Static.generateSite Static.serveStaticFiles cfgPath logPath
+  withGlobalLogging logCfg $ smpServerCLI_ Static.generateSite Static.serveStaticFiles Static.attachStaticFiles cfgPath logPath

apps/smp-server/static/index.html

@@ -221,6 +221,10 @@
               Public information
             </h2>
             <table id="public-info">
+              <tr class="text-grey-black dark:text-white text-base">
+                <td>Server version:</td>
+                <td>${version}</td>
+              </tr>
               <tr class="text-grey-black dark:text-white text-base">
                 <td>Source code:</td>
                 <td><a href="${sourceCode}" target="_blank">${sourceCode}</a></td>
@@ -291,6 +295,12 @@
                   <td>${hostingEntity} (${hostingCountry})</td>
                 </tr>
               </x-hosting>
+              <x-hostingType>
+                <tr class="text-grey-black dark:text-white text-base">
+                  <td>Hosting type:</td>
+                  <td>${hostingType}</td>
+                </tr>
+              </x-hostingType>
               <x-serverCountry>
                 <tr class="text-grey-black dark:text-white text-base">
                   <td>Server country:</td>

apps/smp-server/web/Static.hs

@@ -7,22 +7,30 @@ module Static where
 import Control.Logger.Simple
 import Control.Monad
 import Data.ByteString (ByteString)
-import qualified Data.ByteString as B
+import qualified Data.ByteString.Char8 as B
+import Data.Char (toUpper)
+import Data.IORef (readIORef)
 import Data.Maybe (fromMaybe)
 import Data.String (fromString)
 import Data.Text.Encoding (encodeUtf8)
-import Network.Wai.Application.Static as S
-import Network.Wai.Handler.Warp as W
-import qualified Network.Wai.Handler.WarpTLS as W
+import Network.Socket (getPeerName)
+import Network.Wai (Application)
+import qualified Network.Wai.Application.Static as S
+import qualified Network.Wai.Handler.Warp as W
+import qualified Network.Wai.Handler.Warp.Internal as WI
+import qualified Network.Wai.Handler.WarpTLS as WT
 import Simplex.Messaging.Encoding.String (strEncode)
+import Simplex.Messaging.Server (AttachHTTP)
 import Simplex.Messaging.Server.Information
 import Simplex.Messaging.Server.Main (EmbeddedWebParams (..), WebHttpsParams (..))
+import Simplex.Messaging.Transport (simplexMQVersion)
 import Simplex.Messaging.Transport.Client (TransportHost (..))
 import Simplex.Messaging.Util (tshow)
 import Static.Embedded as E
 import System.Directory (createDirectoryIfMissing)
 import System.FilePath
 import UnliftIO.Concurrent (forkFinally)
+import UnliftIO.Exception (bracket, finally)
 
 serveStaticFiles :: EmbeddedWebParams -> IO ()
 serveStaticFiles EmbeddedWebParams {webStaticPath, webHttpPort, webHttpsParams} = do
@@ -31,9 +39,44 @@ serveStaticFiles EmbeddedWebParams {webStaticPath, webHttpPort, webHttpsParams}
     W.runSettings (mkSettings port) (S.staticApp $ S.defaultFileServerSettings webStaticPath)
   forM_ webHttpsParams $ \WebHttpsParams {port, cert, key} -> flip forkFinally (\e -> logError $ "HTTPS server crashed: " <> tshow e) $ do
     logInfo $ "Serving static site on port " <> tshow port <> " (TLS)"
-    W.runTLS (W.tlsSettings cert key) (mkSettings port) (S.staticApp $ S.defaultFileServerSettings webStaticPath)
+    WT.runTLS (WT.tlsSettings cert key) (mkSettings port) app
   where
-    mkSettings port = setPort port defaultSettings
+    app = staticFiles webStaticPath
+    mkSettings port = W.setPort port warpSettings
+
+-- | Prepare context and prepare HTTP handler for TLS connections that already passed TLS.handshake and ALPN check.
+attachStaticFiles :: FilePath -> (AttachHTTP -> IO ()) -> IO ()
+attachStaticFiles path action =
+  -- Initialize global internal state for http server.
+  WI.withII warpSettings $ \ii -> do
+    action $ \socket cxt -> do
+      -- Initialize internal per-connection resources.
+      addr <- getPeerName socket
+      withConnection addr cxt $ \(conn, transport) ->
+        withTimeout ii conn $ \th ->
+          -- Run Warp connection handler to process HTTP requests for static files.
+          WI.serveConnection conn ii th addr transport warpSettings app
+  where
+    app = staticFiles path
+    -- from warp-tls
+    withConnection socket cxt = bracket (WT.attachConn socket cxt) (terminate . fst)
+    -- from warp
+    withTimeout ii conn =
+      bracket
+        (WI.registerKillThread (WI.timeoutManager ii) (WI.connClose conn))
+        WI.cancel
+    -- shared clean up
+    terminate conn = WI.connClose conn `finally` (readIORef (WI.connWriteBuffer conn) >>= WI.bufFree)
+
+warpSettings :: W.Settings
+warpSettings = W.setGracefulShutdownTimeout (Just 1) W.defaultSettings
+
+staticFiles :: FilePath -> Application
+staticFiles root = S.staticApp settings
+  where
+    settings = (S.defaultFileServerSettings root)
+      { S.ssListing = Nothing
+      }
 
 generateSite :: ServerInformation -> Maybe TransportHost -> FilePath -> IO ()
 generateSite si onionHost sitePath = do
@@ -78,6 +121,7 @@ serverInformation ServerInformation {config, information} onionHost = render E.i
       where
         basic =
           [ ("sourceCode", Just . encodeUtf8 $ sourceCode spi),
+            ("version", Just $ B.pack simplexMQVersion),
             ("website", encodeUtf8 <$> website spi)
           ]
         conds ServerConditions {conditions, amendments} =
@@ -109,7 +153,8 @@ serverInformation ServerInformation {config, information} onionHost = render E.i
             ("hostingCountry", encodeUtf8 <$> country)
           ]
         server =
-          [ ("serverCountry", fmap encodeUtf8 $ serverCountry =<< information)
+          [ ("serverCountry", encodeUtf8 <$> serverCountry spi),
+            ("hostingType",  (\s -> maybe s (\(c, rest) -> toUpper c `B.cons` rest) $ B.uncons s) . strEncode <$> hostingType spi)
           ]
 
 -- Copy-pasted from simplex-chat Simplex.Chat.Types.Preferences

cabal.project

@@ -4,6 +4,14 @@ packages: .
 -- packages: . ../http2
 -- packages: . ../network-transport
 
+-- package *
+   -- coverage: True
+   -- library-coverage: True
+
+-- package attoparsec
+   -- coverage: False
+   -- library-coverage: False
+
 index-state: 2023-12-12T00:00:00Z
 
 package cryptostore
@@ -28,3 +36,17 @@ source-repository-package
     type: git
     location: https://github.com/simplex-chat/sqlcipher-simple.git
     tag: a46bd361a19376c5211f1058908fc0ae6bf42446
+
+-- waiting for published warp-tls-3.4.7
+source-repository-package
+    type: git
+    location: https://github.com/yesodweb/wai.git
+    tag: ec5e017d896a78e787a5acea62b37a4e677dec2e
+    subdir: warp-tls
+
+-- backported fork due http-5.0
+source-repository-package
+    type: git
+    location: https://github.com/simplex-chat/wai.git
+    tag: 2f6e5aa5f05ba9140ac99e195ee647b4f7d926b0
+    subdir: warp

install.sh

@@ -2,10 +2,6 @@
 set -eu
 
 # Links to scripts/configs
-bin="https://github.com/simplex-chat/simplexmq/releases/latest/download"
-bin_smp="$bin/smp-server-ubuntu-20_04-x86-64"
-bin_xftp="$bin/xftp-server-ubuntu-20_04-x86-64"
-
 scripts="https://raw.githubusercontent.com/simplex-chat/simplexmq/stable/scripts/main"
 scripts_systemd_smp="$scripts/smp-server.service"
 scripts_systemd_xftp="$scripts/xftp-server.service"
@@ -26,6 +22,8 @@ path_conf_var="/var/opt"
 path_conf_smp="$path_conf_etc/simplex $path_conf_var/simplex"
 path_conf_xftp="$path_conf_etc/simplex-xftp $path_conf_var/simplex-xftp /srv/xftp"
 
+path_conf_info="$path_conf_etc/simplex-info"
+
 path_systemd="/etc/systemd/system"
 path_systemd_smp="$path_systemd/smp-server.service"
 path_systemd_xftp="$path_systemd/xftp-server.service"
@@ -56,7 +54,7 @@ ${GRN}1.${NC} Install latest binaries from GitHub releases:
 ${GRN}2.${NC} Create server directories:
     - smp: ${YLW}${path_conf_smp}${NC}
     - xftp: ${YLW}${path_conf_xftp}${NC}
-${GRN}3.${NC} Setup user for each server:
+${GRN}3.${NC} Setup user for server:
     - xmp: ${YLW}${user_smp}${NC}
     - xftp: ${YLW}${user_xftp}${NC}
 ${GRN}4.${NC} Create systemd services:
@@ -65,7 +63,12 @@ ${GRN}4.${NC} Create systemd services:
 ${GRN}5.${NC} Install stopscript (systemd), update and uninstallation script:
     - all: ${YLW}${path_bin_update}${NC}, ${YLW}${path_bin_uninstall}${NC}, ${YLW}${path_bin_stopscript}${NC}
 
-Press ${GRN}ENTER${NC} to continue or ${RED}Ctrl+C${NC} to cancel installation"
+Press:
+  - ${GRN}1${NC} to install smp server
+  - ${GRN}2${NC} to install xftp server
+  - ${RED}Ctrl+C${NC} to cancel installation
+  
+Selection: "
 
 end="Installtion is complete!
 
@@ -76,27 +79,79 @@ Please checkout our server guides:
 To uninstall with full clean-up, simply run: ${YLW}sudo /usr/local/bin/simplex-servers-uninstall${NC}
 "
 
+set_version() {
+  ver="${VER:-latest}"
+
+  case "$ver" in 
+    latest)
+      bin="https://github.com/simplex-chat/simplexmq/releases/latest/download"
+      remote_version="$(curl --proto '=https' --tlsv1.2 -sSf -L https://api.github.com/repos/simplex-chat/simplexmq/releases/latest | grep -i "tag_name" | awk -F \" '{print $4}')"
+      ;;
+    *)
+      bin="https://github.com/simplex-chat/simplexmq/releases/download/${ver}"
+      remote_version="${ver}"
+      ;;
+  esac
+}
+
+os_test() {
+ . /etc/os-release
+
+ case "$VERSION_ID" in
+  20.04|22.04) : ;;
+  24.04) VERSION_ID='22.04' ;;
+  *) printf "${RED}Unsupported Ubuntu version!${NC}\nPlease file Github issue with request to support Ubuntu %s: https://github.com/simplex-chat/simplexmq/issues/new\n" "$VERSION_ID" && exit 1 ;;
+ esac
+
+ version="$(printf '%s' "$VERSION_ID" | tr '.' '_')"
+ arch="$(uname -p)"
+
+ case "$arch" in
+  x86_64) arch="$(printf '%s' "$arch" | tr '_' '-')" ;;
+  *)  printf "${RED}Unsupported architecture!${NC}\nPlease file Github issue with request to support %s architecture: https://github.com/simplex-chat/simplexmq/issues/new" "$arch" && exit 1 ;;
+ esac
+
+ bin_smp="$bin/smp-server-ubuntu-${version}-${arch}"
+ bin_xftp="$bin/xftp-server-ubuntu-${version}-${arch}"
+}
+
 setup_bins() {
- curl --proto '=https' --tlsv1.2 -sSf -L "$bin_smp" -o "$path_bin_smp" && chmod +x "$path_bin_smp"
- curl --proto '=https' --tlsv1.2 -sSf -L "$bin_xftp" -o "$path_bin_xftp" && chmod +x "$path_bin_xftp"
+ eval "bin=\$bin_${1}"
+ eval "path=\$path_bin_${1}"
+
+ curl --proto '=https' --tlsv1.2 -sSf -L "$bin" -o "$path" && chmod +x "$path"
+
+ unset bin path
 }
 
 setup_users() {
- useradd -M "$user_smp" 2> /dev/null || true
- useradd -M "$user_xftp" 2> /dev/null || true
+ eval "user=\$user_${1}"
+
+ useradd -M "$user" 2> /dev/null || true
+
+ unset user
 }
 
 setup_dirs() {
  # Unquoted varibles, so field splitting can occur
- mkdir -p $path_conf_smp
- chown "$user_smp":"$user_smp" $path_conf_smp
- mkdir -p $path_conf_xftp
- chown "$user_xftp":"$user_xftp" $path_conf_xftp
+ eval "path_conf=\$path_conf_${1}"
+ eval "user=\$user_${1}"
+
+ mkdir -p $path_conf
+ mkdir -p $path_conf_info
+ printf "local_version_%s='%s'\n" "$1" "$remote_version" >> "$path_conf_info/release"
+ chown -R "$user":"$user" $path_conf
+
+ unset path_conf user
 }
 
 setup_systemd() {
- curl --proto '=https' --tlsv1.2 -sSf -L "$scripts_systemd_smp" -o "$path_systemd_smp"
- curl --proto '=https' --tlsv1.2 -sSf -L "$scripts_systemd_xftp" -o "$path_systemd_xftp"
+ eval "scripts_systemd=\$scripts_systemd_${1}"
+ eval "path_systemd=\$path_systemd_${1}"
+
+ curl --proto '=https' --tlsv1.2 -sSf -L "$scripts_systemd" -o "$path_systemd"
+
+ unset scripts_systemd path_systemd
 }
 
 setup_scripts() {
@@ -110,32 +165,61 @@ checks() {
   printf "This script is intended to be run with root privileges. Please re-run script using sudo."
   exit 1
  fi
+ 
+ set_version
+ os_test
+
+ mkdir -p $path_conf_info  
 }
 
 main() {
  checks
  
- printf "%b\n%b\n" "${BLU}$logo${NC}" "$welcome"
+ printf "%b\n%b" "${BLU}$logo${NC}" "$welcome"
  read ans
 
+ case "$ans" in
+  1) setup='smp' ;;
+  2) setup='xftp' ;;
+  *) printf 'Installation aborted.\n' && exit 0 ;;
+ esac
+
  printf "Installing binaries..."
- setup_bins
+ 
+ for i in $setup; do
+  setup_bins "$i"
+ done
+ 
  printf "${GRN} Done!${NC}\n"
 
  printf "Creating users..."
- setup_users
+
+ for i in $setup; do
+  setup_users "$i"
+ done
+
  printf "${GRN} Done!${NC}\n"
  
  printf "Creating directories..."
- setup_dirs
+
+ for i in $setup; do
+  setup_dirs "$i"
+ done
+
  printf "${GRN} Done!${NC}\n"
 
  printf "Creating systemd services..."
- setup_systemd
+
+ for i in $setup; do
+  setup_systemd "$i"
+ done
+
  printf "${GRN} Done!${NC}\n"
 
  printf "Installing stopscript, update and uninstallation script..."
+
  setup_scripts
+
  printf "${GRN} Done!${NC}\n"
 
  printf "%b" "$end"

package.yaml

@@ -1,211 +0,0 @@
-name: simplexmq
-version: 5.8.2.0
-synopsis: SimpleXMQ message broker
-description: |
-  This package includes <./docs/Simplex-Messaging-Server.html server>,
-  <./docs/Simplex-Messaging-Client.html client> and
-  <./docs/Simplex-Messaging-Agent.html agent> for SMP protocols:
-  .
-  * <https://github.com/simplex-chat/simplexmq/blob/master/protocol/simplex-messaging.md SMP protocol>
-  * <https://github.com/simplex-chat/simplexmq/blob/master/protocol/agent-protocol.md SMP agent protocol>
-  .
-  See <https://github.com/simplex-chat/simplex-chat terminal chat prototype> built with SimpleXMQ broker.
-
-homepage: https://github.com/simplex-chat/simplexmq#readme
-license: AGPL-3
-author: simplex.chat
-maintainer: chat@simplex.chat
-copyright: 2020-2022 simplex.chat
-category: Chat, Network, Web, System, Cryptography
-extra-source-files:
-  - README.md
-  - CHANGELOG.md
-  - cbits/sha512.h
-  - cbits/sntrup761.h
-  - apps/smp-server/static/*.html
-  - apps/smp-server/static/media/*
-
-dependencies:
-  - aeson == 2.2.*
-  - ansi-terminal >= 0.10 && < 0.12
-  - asn1-encoding == 0.9.*
-  - asn1-types == 0.3.*
-  - async == 2.2.*
-  - attoparsec == 0.14.*
-  - base >= 4.14 && < 5
-  - base64-bytestring >= 1.0 && < 1.3
-  - case-insensitive == 1.2.*
-  - composition == 1.0.*
-  - constraints >= 0.12 && < 0.14
-  - containers == 0.6.*
-  - crypton == 0.34.*
-  - crypton-x509 == 1.7.*
-  - crypton-x509-store == 1.6.*
-  - crypton-x509-validation == 1.6.*
-  - cryptostore == 0.3.*
-  - data-default == 0.7.*
-  - direct-sqlcipher == 2.3.*
-  - directory == 1.3.*
-  - filepath == 1.4.*
-  - hourglass == 0.2.*
-  - http-types == 0.12.*
-  - http2 >= 4.2.2 && < 4.3
-  - ini == 0.4.1
-  - iproute == 1.7.*
-  - iso8601-time == 0.1.*
-  - memory == 0.18.*
-  - mtl >= 2.3.1 && < 3.0
-  - network >= 3.1.2.7 && < 3.2
-  - network-info >= 0.2 && < 0.3
-  - network-transport == 0.5.6
-  - network-udp >= 0.0 && < 0.1
-  - optparse-applicative >= 0.15 && < 0.17
-  - process == 1.6.*
-  - random >= 1.1 && < 1.3
-  - simple-logger == 0.1.*
-  - socks == 0.6.*
-  - sqlcipher-simple == 0.4.*
-  - stm == 2.5.*
-  - temporary == 1.3.*
-  - time == 1.12.*
-  - time-manager == 0.0.*
-  - tls >= 1.7.0 && < 1.8
-  - transformers == 0.6.*
-  - unliftio == 0.2.*
-  - unliftio-core == 0.2.*
-  - websockets == 0.12.*
-  - yaml == 0.11.*
-  - zstd == 0.1.3.*
-
-flags:
-  swift:
-    description: Enable swift JSON format
-    manual: True
-    default: False
-  use_crypton:
-    description: Use crypton etc. in cryptostore
-    manual: True
-    default: True
-
-# cpp-options:
-#   - -Dslow_servers
-
-when:
-  - condition: flag(swift)
-    cpp-options:
-      - -DswiftJSON
-  - condition: impl(ghc >= 9.6.2)
-    dependencies:
-      - bytestring == 0.11.*
-      - template-haskell == 2.20.*
-      - text >= 2.0.1 && < 2.2
-  - condition: impl(ghc < 9.6.2)
-    dependencies:
-      - bytestring == 0.10.*
-      - template-haskell == 2.16.*
-      - text >= 1.2.3.0 && < 1.3
-
-library:
-  source-dirs: src
-  c-sources:
-    - cbits/sha512.c
-    - cbits/sntrup761.c
-  include-dirs: cbits
-  extra-libraries: crypto
-
-executables:
-  smp-server:
-    source-dirs:
-    - apps/smp-server
-    - apps/smp-server/web
-    main: Main.hs
-    dependencies:
-      - file-embed
-      - simplexmq
-      - wai-app-static
-      - warp
-      - warp-tls
-    ghc-options:
-      - -threaded
-      - -rtsopts
-
-  ntf-server:
-    source-dirs: apps/ntf-server
-    main: Main.hs
-    dependencies:
-      - simplexmq
-    ghc-options:
-      - -threaded
-      - -rtsopts
-
-  xftp-server:
-    source-dirs: apps/xftp-server
-    main: Main.hs
-    dependencies:
-      - simplexmq
-    ghc-options:
-      - -threaded
-      - -rtsopts
-
-  xftp:
-    source-dirs: apps/xftp
-    main: Main.hs
-    dependencies:
-      - simplexmq
-    ghc-options:
-      - -threaded
-      - -rtsopts
-
-tests:
-  simplexmq-test:
-    source-dirs: tests
-    main: Test.hs
-    dependencies:
-      - simplexmq
-      - deepseq == 1.4.*
-      - generic-random == 1.5.*
-      - hspec == 2.11.*
-      - hspec-core == 2.11.*
-      - HUnit == 1.6.*
-      - QuickCheck == 2.14.*
-      - silently == 1.2.*
-      - main-tester == 0.2.*
-      - timeit == 2.0.*
-    ghc-options:
-      - -threaded
-      - -rtsopts
-      - -with-rtsopts=-A64M
-      - -with-rtsopts=-N1
-
-ghc-options:
-  # - -haddock
-  - -Weverything
-  - -Wno-missing-exported-signatures
-  - -Wno-missing-import-lists
-  - -Wno-missed-specialisations
-  - -Wno-all-missed-specialisations
-  - -Wno-unsafe
-  - -Wno-safe
-  - -Wno-missing-local-signatures
-  - -Wno-missing-kind-signatures
-  - -Wno-missing-deriving-strategies
-  - -Wno-monomorphism-restriction
-  - -Wno-prepositive-qualified-module
-  - -Wno-unused-packages
-  - -Wno-implicit-prelude
-  - -Wno-missing-safe-haskell-mode
-  - -Wno-missing-export-lists
-  - -Wno-partial-fields
-  - -Wcompat
-  - -Werror=incomplete-record-updates
-  - -Werror=incomplete-patterns
-  - -Werror=incomplete-uni-patterns
-  - -Werror=missing-methods
-  - -Werror=tabs
-  - -Wredundant-constraints
-  - -Wincomplete-record-updates
-  - -Wunused-type-patterns
-  - -O2
-
-default-extensions:
-  - StrictData

protocol/xftp.md

@@ -559,11 +559,13 @@ In current implementation of XFTP protocol in SimpleX Chat clients don't use FAC
 
  - perform traffic correlation attacks against senders and recipients and correlate senders and recipients within the monitored set, frustrated by the number of users on the servers.
 
- - observe how much traffic is being sent, and make guesses as to its purpose
+ - observe how much traffic is being sent, and make guesses as to its purpose.
 
-*cannot, even in case of a compromised transport protocol:*
+ - in case of a compromised transport protocol, correlate file senders and receivers.
 
- - perform traffic correlation attacks with any increase in efficiency over a non-compromised transport protocol
+*cannot, in case of a non-compromised transport protocol:*
+
+ - perform traffic correlation attacks.
 
 #### XFTP server
 

protocol/xrcp.md

@@ -67,7 +67,7 @@ The session invitation contains this data:
 - CA TLS certificate fingerprint of the controller - this is part of long term identity of the controller established during the first session, and repeated in the subsequent session announcements.
 - Session Ed25519 public key used to verify the announcement and commands - this mitigates the compromise of the long term signature key, as the controller will have to sign each command with this key first.
 - Long-term Ed25519 public key used to verify the announcement and commands - this is part of the long term controller identity.
-- Session X25519 DH key and SNTRUP761 KEM encapsulation key to agree session encryption (both for multicast announcement and for commands and responses in TLS), as described in https://datatracker.ietf.org/doc/draft-josefsson-ntruprime-hybrid/. The new keys are used for each session, and if client key is already available (from the previous session), the computed shared secret will be used to encrypt the announcement multicast packet. The out-of-band invitation is unencrypted. DH public key and KEM encapsulation key are sent unencrypted. NaCL crypto_box is used for encryption.
+- Session X25519 DH key to agree session encryption (both for multicast announcement and for commands and responses in TLS), as described in https://datatracker.ietf.org/doc/draft-josefsson-ntruprime-hybrid/. The new keys are used for each session, and if client key is already available (from the previous session), the computed shared secret will be used to encrypt the announcement multicast packet. The out-of-band invitation is unencrypted. DH public key and KEM encapsulation key are sent unencrypted. NaCL crypto_box is used for encryption.
 
 Host application decrypts (except the first session) and validates the invitation:
 - Session signature is valid.
@@ -184,7 +184,7 @@ The controller decrypts (including the first session) and validates the received
 The controller should reply with with `ctrlHello` or `ctrlError` response:
 
 ```abnf
-ctrlHello = %s"HELLO " kemCiphertext nonce encrypted(unpaddedSize ctrlHelloJSON helloPad) pad
+ctrlHello = %s"HELLO " kemCiphertext encrypted(unpaddedSize ctrlHelloJSON helloPad) pad
 ; ctrlHelloJSON is encrypted with the hybrid secret,
 ; including both previously agreed DH secret and KEM secret from kemCiphertext
 unpaddedSize = largeLength
@@ -206,6 +206,8 @@ JTD schema for the encrypted part of controller HELLO block `ctrlHelloJSON`:
 }
 ```
 
+Controller `hello` block and all subsequent protocol messages are encrypted with the chain keys derived from the hybrid key (see key exchange below) - that is why conntroller hello block does not include nonce. That provides forward secrecy within the XRCP session. Receiving this `hello` block allows host to compute the same hybrid keys and to derive the same chain keys.
+
 Once the controller replies HELLO to the valid host HELLO block, it should stop accepting new TCP connections.
 
 ### Controller/host session operation
@@ -223,10 +225,12 @@ tlsunique channel binding from TLS session MUST be included in commands (include
 The syntax for encrypted command and response body encoding:
 
 ```abnf
-commandBody = encBody sessSignature idSignature [attachment]
-responseBody = encBody [attachment] ; counter must match command
-encBody = nonce encLength32 encrypted(tlsunique counter body)
-attachment = %x01 nonce encLength32 encrypted(attachment)
+commandBody = counter encBody sessSignature idSignature [attachment]
+responseBody = counter encBody [attachment] ; counter must match command
+; counter is placed outside of encrypted body to allow correlating encryption keys
+; with the chain keys (each command and response are encrypted by different keys)
+encBody = encLength32 encrypted(tlsunique body)
+attachment = %x01 encLength32 encrypted(attachment)
 noAttachment = %x00
 tlsunique = length 1*OCTET
 counter = 8*8 OCTET ; int64
@@ -239,7 +243,7 @@ If the command or response includes attachment, its hash must be included in com
 
 Initial announcement is shared out-of-band (URI with xrcp scheme), and it is not encrypted.
 
-This announcement contains only DH keys, as KEM key is too large to include in QR code, which are used to agree encryption key for host HELLO block. The host HELLO block will contain DH key in plaintext part and KEM encapsulation (public) key in encrypted part, that will be used to determine the shared secret (using SHA256 over concatenated DH shared secret and KEM encapsulated secret) both for controller HELLO response (that contains KEM ciphertext in plaintext part) and subsequent session commands and responses.
+This announcement contains only DH keys, as KEM key is too large to include in QR code, which are used to agree encryption key for host HELLO block. The host HELLO block will contain DH key in plaintext part and KEM encapsulation (public) key in encrypted part, that will be used to determine the shared secret (using SHA3-256 over concatenated DH shared secret and KEM encapsulated secret) to derive keys for controller HELLO response (that contains KEM ciphertext in plaintext part) and subsequent session commands and responses.
 
 During the next session the announcement is sent via encrypted multicast block. The shared key for this announcement and for host HELLO block is determined using the KEM shared secret from the previous session and DH shared secret computed using the host DH key from the previous session and the new controller DH key from the announcement.
 
@@ -250,7 +254,7 @@ In pseudo-code:
 ```
 // session 1
 hostHelloSecret(1) = dhSecret(1)
-sessionSecret(1) = sha256(dhSecret(1) || kemSecret(1)) // to encrypt session 1 data, incl. controller hello
+sessionSecret(1) = sha3-256(dhSecret(1) || kemSecret(1)) // to encrypt session 1 data, incl. controller hello
 dhSecret(1) = dh(hostHelloDhKey(1), controllerInvitationDhKey(1))
 kemCiphertext(1) = enc(kemSecret(1), kemEncKey(1))
 // kemEncKey is included in host HELLO, kemCiphertext - in controller HELLO
@@ -262,7 +266,7 @@ dhSecret(n') = dh(hostHelloDhKey(n - 1), controllerDhKey(n))
 
 // session n
 hostHelloSecret(n) = dhSecret(n)
-sessionSecret(n) = sha256(dhSecret(n) || kemSecret(n)) // to encrypt session n data, incl. controller hello
+sessionSecret(n) = sha3-256(dhSecret(n) || kemSecret(n)) // to encrypt session n data, incl. controller hello
 dhSecret(n) = dh(hostHelloDhKey(n), controllerDhKey(n))
 // controllerDhKey(n) is either from invitation or from multicast announcement
 kemCiphertext(n) = enc(kemSecret(n), kemEncKey(n))
@@ -273,6 +277,22 @@ If controller fails to store the new host DH key after receiving HELLO block, th
 
 To decrypt a multicast announcement, the host should try to decrypt it using the keys of all known (paired) remote controllers.
 
+Once kemSecret is agreed for the session, it is used to derive two chain keys, to receive and to send messages:
+
+```
+host: sndKey, rcvKey = HKDF(kemSecret, "SimpleXSbChainInit", 64)
+controller: rcvKey, sndKey  = HKDF(kemSecret, "SimpleXSbChainInit", 64)
+```
+
+where HKDF is based on SHA512, with empty salt.
+
+Actual keys and nonces to encrypt and decrypt messages are derived from these chain keys:
+
+```
+to send: (sndKey', sk, nonce) = HKDF(sndKey, "SimpleXSbChain", 88)
+to receive: (rcvKey', sk, nonce) = HKDF(rcvKey, "SimpleXSbChain", 88)
+```
+
 ## Threat model
 
 #### A passive network adversary able to monitor the site-local traffic:

rfcs/2024-06-21-short-links.md

@@ -0,0 +1,124 @@
+# Short invitation links
+
+## Problem
+
+Long links look scary and unsafe for many users. While this is a perceived problem, rather than a real one, it hurts adoption.
+
+What is worse, long links do not fit in profile descriptions of other social networks where people might want to advertize their contact addresses.
+
+The current link size limitation is also the reason for not including PQ KEM keys into invitation links and addresses, postponing the moment when PQ-resistant encryption kicks in - if we include PQ KEM key into the link, the QR code will not be scannable.
+
+Additionally, if we store short links, they can also include chat preferences and public profile data.
+
+## Solution
+
+MITM-resistant link shortening.
+
+Instead of generating the random address that would resolve into the link - doing so would create the possibility of MITM by the server hosting this link - we can use private key as the link ID that will be passed to the accepting party, and the hash of the public key as ID for the server - the accepting party would present this key itself as ID and it will also be used for server to client encryption (see Protocol below). HKDF will be used to derive symmetric key from private key and used in secret_box together with random nonce (to allow replacing data with the same key but with a different nonce - nonce will be sent to the server too). secret_box construction is authenticated encryption, so it would protect from MITM.
+
+The proposed syntax:
+
+```abnf
+shortConnectionRequest = connectionScheme "/" connReqType "#/" smpServer "/" linkHash
+connReqType = %s"invitation" / %s"contact"
+connectionScheme = (%s"https://" clientAppServer) / %s"simplex:"
+clientAppServer = hostname [ ":" port ]
+  ; client app server, e.g. simplex.chat
+smpServer = serverIdentity "@" srvHosts [":" port] ; no smp:// prefix, no escaping
+srvHosts = <hostname> ["," srvHosts] ; RFC1123, RFC5891
+linkHash = <base64url encoded SHA256 or SHA512 hash of the original link>
+```
+
+If SMP server supports pages, its name can be used as clientAppServer, without repeating it after #, for a shorter link.
+
+Example link:
+
+```
+https://simplex.chat/contact/#0YuTwO05YJWS8rkjn9eLJDjQhFKvIYd8d4xG8X1blIU=@smp8.simplex.im/abcdefghij0123456789abcdefghij0123456789abc=
+```
+
+This link has the length of ~136 characters (256 bits), which is shorter than the full contact address (~310 characters) and much shorter than invitation links (~528 characters) even without post-quantum keys added to them.
+
+This size can be further reduced by
+- use server domain in the link.
+- do not include onion address, as the connection happens via proxy anyway, if it's untrusted server.
+- not pinning server TLS certificate - the downside here is that while the attack that compromises TLS will not be able to substitute the link (because it's hash will not match), it will be able to intercept and to block it.
+- using shorter hash, e.g. SHA128 - reducing the collision resistance.
+
+If the server is known, the client could use it's hash and onion address, otherwise it could trust the proxy to use any existing session with the same hostname or to accept the risk of interception - given that there is no risk of substitution.
+
+With the first two of these "improvements" the link could be ~122 characters:
+
+```
+https://smp8.simplex.im/contact/#0YuTwO05YJWS8rkjn9eLJDjQhFKvIYd8d4xG8X1blIU@/abcdefghij0123456789abcdefghij0123456789abc
+```
+
+If onion address is preserved the link will be ~184 characters (won't fit in Twitter 160 characters bio):
+
+```
+https://smp8.simplex.im/contact/#0YuTwO05YJWS8rkjn9eLJDjQhFKvIYd8d4xG8X1blIU@beccx4yfxxbvyhqypaavemqurytl6hozr47wfc7uuecacjqdvwpw2xid.onion/abcdefghij0123456789abcdefghij0123456789abc
+```
+
+If we implement it, the request to resolve the link would be made via proxied SMP command (to avoid the direct connection between the client and the recipient's server).
+
+Pros:
+- a bit shorter link.
+- possibility to include post-quantum keys into the full link keeping the same shortened link size.
+- possibility to include chat profile of contact or group, and preferences, for a much better connection experience, and to show this information when the link sent in the conversation (clients can resolve them automatically, without connecting - it can be resolved by the sending clients).
+- server will not have access to the link.
+
+Cons:
+- protocol complexity.
+- observers can access the link content, so for 1-time invitation we should only include permissions and not profile.
+
+Pros are a huge improvement of UX of connecting both within and from outside of the app (e.g., link can be resolved even before creating chat profile, as part of the onboarding).
+
+## Protocol
+
+To support short links, the SMP servers would provide a simple key-value store enabled by three additional commands: `WRT`, `CLR` and `READ`
+
+`WRT` command is used to store and to update values in the store. The size of the value is limited by the same size as sent messages (or, possibly, smaller - as connection information size used in confirmation messages) - the clients would use this fixed size irrespective of the content. `WRT` command will be sent with the data blob ID in the transaction entityId field, public authorization key used to authorize `WRT` and `CLR` commands (subsequent WRT commands to the existing key must use the same key), and the data blob.
+
+`CLR` command must use with the same entity ID and must be authorized by the same key.
+
+`READ` command must use the ID which hash would be equal of the ID used to create the data blob, and this ID would also be used as public authorization
+
+## Algorithm to store and to retrieve data blob.
+
+**Store data blob**
+
+- the data blob owner generates X25519 key pair: `(k, pk)`.
+- private key `pk` will be included in the short link shared with the other party (only base64url encoded key bytes, not X509 encoding).
+- `HKDF(pk)` will be used to encrypt the link data with secret_box before storing it on the server.
+- the hash of public key `sha256(k)` will be used as ID by the owner to store and to remove the data blob (`WRT` and `CLR` commands).
+
+**Retrieve data blob**
+
+- the sender uses the public key `k` derived from the private key `pk` included in the link as entity ID to retrieve data blob (the server will compute the ID used by the owner as `sha256(k)` and will be able to look it up). This provides the quality that the traffic of the parties has no shared IDs inside TLS. It also means that unlike message queue creation, the ID to retrieve the blob was never sent to the blob creator, and also is not known to the server in advance (the second part is only an observation, in itself it does not increase security, as server has access to an encrypted blob anyway).
+- note that the sender does not authorize the request to retrieve the blob, as it would not increase security unless a different key is used to authorize, and adding a key would increase link size.
+- server session keys with the sender will be `(sk, spk)`, where `sk` is public key shared with the sender during session handshake, and `spk` is the private key known only to the server.
+- this public key `k` will also be combined with server session key `spk` using `dh(k, spk)` to encrypt the response, so that there is no ciphertext in common in sent and received traffic for these blobs. Correlation ID will be used as a nonce for this encryption.
+- having received the blob, the client can now decrypt it using secret_box with `HKDF(pk)`.
+
+Using the same key as ID for the request, and also to additionally encrypt the response allows to use a single key in the link, without increasing the link size.
+
+## Threat model
+
+**Compromised SMP server**
+
+can:
+- delete link data.
+- hide link selectively from some requests.
+
+cannot:
+- undetectably replace link data.
+- access unencrypted link data, whether it was or was not accessed by the accepting party.
+- observe IP addresses of the users accessing link data.
+
+**Passive observer who observed short link**:
+
+can:
+- access original unencrypted link data
+
+cannot:
+- replace or delete the link data

rfcs/2024-07-06-ios-notifications.md

@@ -0,0 +1,18 @@
+# iOS notifications stability
+
+## Problem
+
+iOS notifications may fail to deliver for several reasons, but there are two important reasons that we could address:
+- when notification server is not subscribed to SMP server(s), the notifications can be dropped - it can happen because either notification server restarts or becuase SMP server restarted and some messages are received before notification server resubscribed. We lose approximately 3% of notifications because of this reason.
+- when user device is offline or has low power condition, Apple does not deliver notification, but puts them to storage. If while the notification is in storage a new one arrives it would overwrite the previous notification. If it was the message to the same message queue, the client will download messages anyway, up to a limit, but if the message was to another queue, it will not be delivered until the app is opened. Apple delivers about 88% of notifications that should be delivered (not accounting for uninstalled apps), the rest is replaced with the newer notifications.
+
+## Solution
+
+The first problem can be solved by preserving notifications for a limited time (say 1 hour) in case there is no subscription to notification from notification server. At the very least, they can be preserved in SMP server memory but can also be stored to a file on restart, similar to messages, and be delivered when notification server resubscribes. It is sufficient to store one notification per messaging queue.
+
+The second problem is both more damaging and more complex to solve. The solution could be to always deliver several last notifications to different queues in one packet (Apple allows up to ~4-5kb notification size, and we are sending packets of fixed size 512 bytes, so we could fit up to 8-10 of them in each notification).
+
+Every time a client receives such batch of notifications if can:
+- check if that notification was already received in the previous batch.
+- if it was received, it would be ignored, otherwise it would be processed.
+- process them one by one, started from the most recent one while the time allows.

rfcs/2024-09-01-smp-message-storage.md

@@ -0,0 +1,179 @@
+# SMP server message storage
+
+## Problem
+
+Currently SMP servers store all queues in server memory. As the traffic grows, so does the number of undelivered messages. What is worse, Haskell is not avoiding heap fragmentation when messages are allocated and then de-allocated - undelivered messages use ByteString and GC cannot move them around, as they use pinned memory.
+
+## Possible solutions
+
+### Solution 1: solve only GC fragmentation problem
+
+Move from ByteString to some other primitive to store messages in memory long term, e.g. ShortByteString, or manage allocation/de-allocation of stored messages manually in some other way.
+
+Pros: the simplest solution that avoids substantial re-engineering of the server.
+
+Cons:
+- not a long term solution, as memory growth still has limits.
+- may be ineffective, as it introduces additional copying of bytes.
+
+### Solution 2: move message storage to hard drive
+
+Use files or RocksDB to store messages.
+
+Pros:
+- much lower memory usage.
+- no message loss in case of abnormal server termination (important until clients have delivery redundancy).
+- this is a long term solution, and at some point it might need to be done anyway.
+
+Cons:
+- substantial re-engineering costs and risks.
+- metadata privacy. Currently we only save undelivered messages when server is restarted, with this approach all messages will be stored for some time. this argument is limited, as hosting providers of VMs can make memory snapshots too, on the other hand they are harder to analyze than files. On another hand, with this approach messages will be stored for a shorter time.
+
+#### RocksDB and other key-value stores
+
+The downside of any key-value stores is that they don't seem to have efficient primitives for sequential delivery. While sequential delivery can be modelled with linked lists, they would require 1 key insert (on send), 3 key updates (1 update to update queue data on send, 1 update of the last message to point to the next, 1 update on delivery or message expiration) and 1 key deletion (on delivery or message expiration) for each delivered message.
+
+This might result in substantial write amplification and compacting costs.
+
+In general, tree structures that are efficient for quick lookups and updates, given approximately fixed value size, are inefficient for modelling queues.
+
+#### Files
+
+The upside of files is that they are well suited for sequential delivery and don't result in the same churn, with careful design, as trees do.
+
+The downside of filesystem is that it does not scale well with the large number of files in a folder, so queues will have to be spread across multiple folders, following tree-like structure.
+
+I could not find an available library that efficiently models sequential delivery in highly concurrent environment.
+
+A possible design could be the following.
+
+##### Queue folder and files
+
+Each message queue is stored in its own folder (see below on folder locations). Folder would contain these files:
+
+- `messages.abcd.log` - the file that is used to read messages from, sequentially
+- `messages.efgh.log` - the optional file that is used to write messages, in case it is different from read file.
+- `queue.log` - append-only file where the last line represents the current queue state
+- `queue.timestamp.log` - previous states of queue.log file
+
+Each line in "queue.log" file has this syntax
+
+```abnf
+queueLogLine =
+    %s"read_file=" base64
+    %s"read_msg=" digits
+    %s"read_byte=" digits
+    %s"write_file=" base64
+    %s"write_msg=" digits
+```
+
+When queue is first requested by the server:
+
+```c
+if queue folder exists:
+    read queue state from last line of queue.log
+    if queue.log contained more than one line: // compaction
+        copy queue.log to queue.timestamp.log
+        write one line queue state to queue.log
+else:
+    create queue folder
+    create messages.abcd.log (abcd is some random string)
+    read_msg = 0
+    read_byte = 0
+    create queue.log with one line: "read_file=abcd read_msg=0 read_byte=0 write_files=abcd write_msg=0"
+open read_file in ReadMode and seek to read_byte position
+nextReadByte = read_byte
+nextReadMsg = read_msg
+open write_file in AppendMode
+```
+
+When message is added to the queue (assumes that queue state is loaded to server memory, if not the previous section will be done first):
+
+```c
+if write_msg > max_queue_messages:
+    return quota error
+else if write_msg = max_queue_messages:
+    add quota_exceeded message to write_file
+    update queue state: write_msg += 1
+    append updated queue state to queue.log
+else
+    // It is required that `max_queue_messages < max_file_messages`,
+    // so that we never need more than one additional write file.
+    if write_msg >= max_file_messages: // queue file rotation
+        create messages.efgh.log // efgh is some random string
+        update queue state: write_file=efgh write_msg=0 // read file remains the same as it was
+        append updated queue state to queue.log
+        copy queue.log to queue.timestamp.log
+        // `old` needs to be defined to limit the number and storage duration,
+        // preserving not more than N files, and not more than M days files, "and then some"
+        // (that is if the queue has high churn, we have file from M days before in any case, for any debugging).
+        delete `old` `queue.timestamp.log` files
+        write one line queue state to queue.log // compaction
+
+    add message to write_file
+    update queue state: write_msg += 1
+    append updated queue state to queue.log
+```
+
+The above algorithm assumes `max_queue_messages < than max_file_messages`, so that we never need more than one write file.
+
+When message is delivered, it is simply read from the read queue, queue state does not change yet:
+
+```c
+if nextReadMsg > read_msg:
+    deliver cached message, no need to read it again
+else
+    read message from read_file handle
+    nextReadMsg = read_msg + 1
+    nextReadByte = current position in file
+```
+
+When message delivery is acknowledged, the read queue needs to be advanced, and possibly switched to read from the current write_queue:
+
+```c
+if nextReadByte == read_byte:
+    return error // nothing was delivered
+else if nextReadByte = EOF:
+    // end of file is reached, possibly some other condition,
+    // but it should allow changing max_file_messages on server restart
+    currReadFile = read_file
+    read_file = write_file
+    read_msg = 0
+    read_byte = 0
+    append updated queue state to queue.log
+    delete currReadFile
+else
+    read_msg += 1
+    read_byte = nextReadByte
+    // `seek` should not be necessary, as the handle is already at nextReadByte position here
+    // seek to read_byte
+    append updated queue state to queue.log
+```
+
+The above algorithm delegates the problem of compaction and fragmentation management to file system, that is very optimized for such scenarios.
+
+Also, read and write files will grow to almost a constant size, so the space they used may be re-used.
+
+An important consideration is that writes to queue.log and message.log files and queue state modifications have to be sequential, without concurrency - it can be managed with the usual locks.
+
+##### Queue folders structure
+
+Most Linux systems use EXT4 filesystem where the file lookup time scales linearly to the number of files. While alternatives with logarithmic lookup time exist (XFS), they may be very complex to configure on the existing systems.
+
+So storing all queue folders in one folder won't scale.
+
+To solve this problem we could use recipient queue ID in base64url format not as a folder name, but as a folder path, splitting it to path fragments of some length. The number of fragments can be configurable and migration to a different fragment size can be supported as the number of queues on a given server grows.
+
+Currently, queue ID is 24 bytes random number, thus allowing 2^192 possible queue IDs. If we assume that a server must hold 1b queues, it means that we have ~2^162 possible addresses for each existing queue. 24 bytes in base64 is 32 characters that can be split into say 8 fragments with 4 characters each, so that queue folder path for queue with ID `abcdefghijklmnopqrstuvwxyz012345` would be:
+
+`/var/opt/simplex/messages/abcd/efgh/ijkl/mnop/qrst/uvwx/yz01/2345`
+
+The maximum theoretic number of the folders on the 1st level is 64^4, or 2^24 ~ 16m - this is probably still a large number of subfolders for EXT4. Given that addresses are random, all the possible combinations in the first folder can be used with a large number of queues.
+
+So we could use an unequal split of path, two letters each and the last being long:
+
+`/var/opt/simplex/messages/ab/cd/ef/ghijklmnopqrstuvwxyz012345`
+
+The first three levels in this case can have 4096 subfolders each, and it gives 68b possible subfolders (64^2^3), so the last level will be sparse in case of 1b queues on the server. So we could make it 4 levels with 2 letters to never think about it, accounting for a large variance of the random numbers distribution:
+
+`/var/opt/simplex/messages/ab/cd/ef/gh/ijklmnopqrstuvwxyz012345`

rfcs/2024-09-05-queue-storage.md

@@ -0,0 +1,81 @@
+# Storage considerations for SMP queues
+
+See [Short invitation links](./2024-06-21-short-links.md).
+
+## Problem
+
+1) queue records are created permanently, until the clients delete them.
+
+2) clients only delete queue records based on some user action, pending connections do not expire.
+
+While part 2 should be improved in the client, indefinite storage of queue records becomes a much bigger issue if each of them would result in a permanent storage of 4-16kb blob in server memory, without server-side expiration for short invitation links.
+
+## Possible solutions
+
+1) Add some queue timestamp, e.g. queue creation date, to expire unsecured queues after say 3 weeks.
+
+The problem with this approach is that contact addresses are also unsecured queues, and they should not be expired.
+
+We could set really large expiration time, and require that clients "update" the unsecured queues they need at least every 1-2 years, but it would not solve the problem of storing a large number of blobs in the server memory for unused/abandoned 1-time invitations.
+
+2) Do not store blobs in memory / append-only log, and instead use something like RocksDB. While it may be a correct long term solution, it may be not expedient enough at the current POC stage for this feature. Also, the lack of expiration is wrong in any case and would indefinitely grow server storage.
+
+3) Add flag allowing the server to differentiate permanent queues used as contact addresses, also using different blob sizes for them. In this case, messaging queues will be expired if not secured after 3 weeks, and contact address queues would be expired if not "updated" by the owner within 2 years.
+
+Probably all three solutions need to be used, to avoid creating a non-expiring blob storage in memory, as in case too many of such blobs are created it would not be possible to differentiate between real users and resource exhaustion attacks, and unlike with messages, they won't be expiring too.
+
+Servers already can differentiate messaging queues and contact address queues, if they want to:
+- with the old 4-message handshake, the confirmation message on a normal queue was different, and also KEY command was eventually used.
+- with the fast 2-message handshake, while the confirmation message has the same syntax, and the differences are inside encrypted envelope, the client still uses SKEY command.
+- in both cases, the usual messaging queues are secured, and contact addresses are not, so this difference is visible in the storage as well (although it is not easy to differentiate between abandoned 1-time invitations and contact addresses).
+
+Differentiating these queues can also allow different message retention times - e.g., the queues for contact addresses could have bigger size, but have lower message retention time.
+
+## Proposed solution
+
+1. Add queue updated_at date into queue records. While it adds some metadata, it seems necessary to manage retention and quality of service. It will not include exact time, only date, and the time of creation will be replaced by the time of any update - queue secured, a message is sent, or queue owner subscribes to the queue. To avoid the need to update store log on every message this information can be appended to store log on server termination. Or given that only one update per day is needed it may be ok to make these updates as they happen (temporarily making the sequence and time of these events available in storage).
+
+2. Add flag to indicate the queue usage - messaging queue or queue for contact address connection requests. This would result in different queue size and different retention policy for queue and its messages. We already have "sender can secure flag" which is, effectively, this flag - contact address queues are never secured. So this does not increase stored metadata in any way.
+
+## Possible changes to short links
+
+This is a design considerations and a concept, not a design yet.
+
+Instead of implementing a generic blob storage that can be used as an attack vector, and adds additional failure point (another server storing blob that is necessary to connect to the queue on the current server), but instead adds an extended queue information blobs, most of which could be dropped without the loss of connectivity, so that the attack can be mitigated by deleting these blobs without users losing the ability to connect, as long as the queue and minimal extended information is retained.
+
+So, to make the connection there need to be these elements:
+
+- queue server and queue ID - mandatory part, that can be included in short link
+- SMP key - mandatory part for all queues. We are considering initializing ratchets earlier for contact addresses, and include ratchet keys and pre-keys into queue data as well, but it is out of scope here.
+- Ratchet keys - mandatory part for 1-time invitation that won't fit in short link.
+- PQ key - optional part that can be stored with addresses if ratchet keys are added and with 1-time invitations.
+- App blobs - chat preferences for 1-time invitation links and profile information for contact addresses.
+
+So rather that storing one blob with a large address inside it, not associated with the queue, increasing probability of failure and reducing our ability to mitigate resource exhaustion, we could store extended blobs associated with the queues.
+
+Also, we need the address shared with the sender (party accepting the connection) to be short. We could use a similar approach that was proposed for data blobs, using a single random seed per queues to derive multiple keys and IDs from it. For example:
+
+1. The queue owner:
+  - generates Ed25529 key pair `(sk, spk)` and X25519 key pair `(dhk, dhpk)` to use with the server, same as now sent in NEW command.
+  - generates queue recipient ID (this ID can still be server-generated).
+  - generates X25519 key pair `(k, pk)` to use with the accepting party.
+  - derives from `k`:
+    - sender ID.
+    - symmetric key for authenticated encryption of blobs.
+  - `k` will be used as short link.
+2. All other data from the invitation can be included in queue creation request and be associated with the queue as 1-3 blobs with different priority:
+  - ratchet keys - it will have a small size, so only this blob cannot be removed, while other blobs can be removed in case of resource exhaustion.
+  - PQ keys - optional blob.
+  - conversation preferences and profile - can be removed depending on creation time, e.g. all new blobs can be removed.
+
+The algorithm used to derive key and ID from `k` needs to be cryptographically secure, e.g. it could be some KDF or ChaCha DRG initialized with `k` as seed, TBC.
+
+So, coupling blob storage with messaging queues has these pros/cons:
+
+Cons:
+- no additional layer of privacy - the server used for connection is visible in the link, even after the blobs are removed from the server.
+
+Pros:
+- no additional point of failure in the connection process - the same server will be used to retrieve necessary blobs as for connection.
+- queue blobs of messaging blobs will be automatically removed once the queue is secured or expired, without additional request from the recipient - reducing the storage and the time these blobs are available.
+- queue blobs for contact addresses will be structured and some of the large blobs can be removed in case of resource exhaustion attack (and recreated by the client if needed), with the only downside that PQ handshake will be postponed (which is the case now) and profile will not be available at a point of connection.

rfcs/2024-09-09-smp-blobs.md

@@ -0,0 +1,80 @@
+# Blob extensions for SMP queues
+
+Evolution of the design for short links, see [here](./2024-06-21-short-links.md) and [here](./2024-09-05-queue-storage.md).
+
+## Problems
+
+Allow storing extended information with SMP queues to improve UX and security of making connections:
+- short invitation links and contact addresses.
+- PQ encryption from the first message.
+- present user profile with chat preferences and welcome message when the public address link is scanned.
+
+## Design
+
+1. Queue creation/update date is already added to server persistence, allowing to expire queues and blobs, depending on their usage.
+2. Add "queue type" metadata to NEW command to indicate whether messaging queue is used as public address or as messaging queue (see previous docs on why it doesn't change threat model). While at the moment it would match sndSecure flag there may be future scenarios when they diverge. Initially only "invitation" and "contact" types will be supported.
+3. Prohibit sndSecure flag for "contact" queues, prohibit securing contact queues.
+4. Add "queue blobs" to NEW command:
+  - blob0: ratchetKeys up to N0 bytes - priority 0, can't be removed by the server, only in "invitation"
+  - blob1: PQ key up to N1 bytes - priority 1, can be removed by the server, only used in "invitation"
+  - blob2: Application data up to N2 bytes - priority 2, can be removed by the server.
+5. Add linkId to NEW command
+6. linkId and blobs will be removed when queue is secured.
+7. Add recipient command to remove/upsert blob2 for contact queues.
+8. Add sender command to retrieve blobs.
+
+## Protocol
+
+### Creating a queue:
+
+The queue owner:
+- generates Ed25529 key pair `(sk, spk)` and X25519 key pair `(dhk, dhpk)` to use with the server, same as now. `sk` and `dhk` will be sent in NEW command.
+- generates X25519 key pair `(k, pk)` to use with the accepting party to encrypt queue messages.
+- derives from `k` using HKDF:
+- symmetric key `bk` for authenticated encryption of blobs.
+- `linkId`, will be sent in NEW command.
+- `k` will be used as short link.
+- sends NEW command.
+
+NEW command syntax:
+
+```abnf
+create = %s"NEW " linkId queueType recipientAuthPublicKey recipientDhPublicKey
+                  basicAuth subscribe sndSecure [ "0" blob0 ] [ "1" blob1 ] [ "2" blob2 ] 
+queueType = %s"I" / %s "C" ; new parameter
+linkId = length *OCTET ; new parameter,
+                        ; can be empty in which case blobs won't be allowed
+blob0 = word16 *OCTET ; new parameter, encrypted ratchet keys,
+                      ; including nonce and auth tag
+blob1 = word16 *OCTET ; new parameter, encrypted PQ key
+blob2 = word16 *OCTET ; new parameter, encrypted application data
+```
+
+SET - command to update queue blobs (recipientId is used as entity ID):
+
+```abnf
+set = %s"SET " linkId [ "2" blob2 ] ; passing empty blob removes it
+linkId ; updated (or the same) linkId, can be empty to remove blobs
+        ; allows to change the address without removing the queue / changing blobs
+        ; (e.g., to avoid losing the messages).
+```
+
+### Sending messages to the queue
+
+GET - command to get queue blobs (linkId is used as entity ID):
+
+```abnf
+get = %s"GET"
+```
+
+Response to GET:
+
+```abnf
+blobs = %s"BLOB" senderId [ "0" blob0 ] [ "1" blob1 ] [ "2" blob2 ]
+```
+
+As blobs are retrieved using a separate linkId, once blobs are removed it will be impossible to find senderId from short link - it is a threat model improvement. Once server storage is compacted, it will be impossible to find queue related to the link even with the access to server data (unless server preserves the data).
+
+### Possible privacy improvement
+
+We could only allow unauthorized GET and authorized SET commands for long-term "contact" queues, and return BLOB in response to SKEY (or require that GET is authorized) - so that only the person who secures the queue will get access to data blobs. This way it ensures that the parties transmitting the invitation links cannot retrieve their content without the sender noticing it.

rfcs/2024-09-10-private-rendezvous.md

@@ -0,0 +1,26 @@
+# Private rendezvous protocol
+
+## Problem
+
+Our current handshake protocol is open to this attack: whoever observes the link exchange, knows on which server connection is being made, and if the traffic on this server is observed, then it can confirm communication between parties. Further, even with the [last proposal](./2024-09-09-smp-blobs.md#possible-privacy-improvement), having real-time access to the server data allows to establish the exact messaging queue that is used to send messages.
+
+## Solution
+
+We could make the initial link exchange more private by making it harder for any observer to discover which server will be used for messaging by hiding this information from the server that hosts the initial link.
+
+Preliminary, the protocol could be the following:
+
+1. Connection initiator stores 224-256 bytes of encrypted connection link on a rendezvous server (link contains server host and linkId on another messaging server, not a rendezvous one).
+
+2. Rendezvous server adds these links to buckets, up to 64 links per bucket. Bucket ID is the timestamp when the bucket was created + a sequential bucket number, in case more than one bucket is created per second.
+
+3. The server responds to the link creator with a bucket ID where this link was added. That bucket ID is its timestamp + a number prevents server "fingerprinting" clients and using say one bucket for each client. If timestamp is different or a bucket number within this timestamp is too large, the client can refuse to use it, depending on the client settings.
+
+4. The initiating party will pass to the accepting party the rendezvous server host, the hash of this bucket ID (bucket link) and the passphrase to derive the key from. The initiating party has an option to pass a link and passphrase via two channels - in which case the link will only contain the bucket ID.
+
+5. The accepting party would then request the bucket via its ID hash (the server would store hashes to be able to look up - hash is used to prevent showing time in the link) and attempt to decrypt all contained links using the provided key.
+The accepting party then will continue the connection via the decrypted link.
+
+This obviously does not protect accepting party from the initiating party, if it can choose rendezvous server it controls. It also does not protect from the malicious rendezvous server that would collaborate with link observers. I think reunion doesn’t protect from it too.
+
+But it does protect connection from whoever observes the link, particularly if this link only contains the bucket and the key is passed separately, via some other channel.

rfcs/2024-09-15-shared-port.md

@@ -0,0 +1,163 @@
+# Sharing protocol ports with HTTPS
+
+Some networks block all ports other than web ports, including port 5223 used for SMP protocol by default. Running SMP servers on a common web port 443 would allow them to work on more networks. The servers would need to provide an HTTPS page for browsers (and probes).
+
+## Problem
+
+Browsers and tools rely on system CA bundles instead of certificate pinning.
+The crypto parameters used by HTTPS are different from what the protocols use.
+Public certificate providers like LetsEncrypt can only sign specific types of keys and Ed25519 isn't one of them.
+
+This means a server should distinguish browser and protocol clients and adjust its behavior to match.
+
+## Solution
+
+`tls` package has a server hook that allows producing a different set of `TLS.Credentials` according to a client-provided "Server Name Indication" extension.
+
+Since LE certificates are only handed out to domain names, TLS client will be sending the SNI.
+However client transports are constructed over connected sockets and the SNI wouldn't be present unless explicitly requested.
+When a client sends SNI, then it's a browser and a web credentials should be used.
+Otherwise it's a protocol client to be offered the self-signed ca, cert and key.
+
+When a transport colocated with a HTTPS, its ALPN list should be extended with `h2 http/1.1`.
+The browsers will send it, and it should be checked before running transport client.
+If HTTP ALPN is detected, then the client connection is served with HTTP `Application` instead (the same "server information" page).
+
+If some client connects to server IP, doesn't send SNI and doesn't send ALPN, it will look like a pre-handshake client.
+In that case a server will send its handshake first.
+This can be mitigated by delaying its handshake and letting the probe to issue its HTTP request.
+
+## Implementation plan
+
+An unmodified client should be able to use protocols on port 443 right away.
+
+The switchover happens inside `runTransportServerState` before `runClient`:
+
+```haskell
+runServer (tcpPort, ATransport t) = do
+  -- ...
+  runTransportServerState_ ss started tcpPort serverParams tCfg $ \socket h -> do -- expose raw socket for warp-tls internals to attach
+    negotiated <- getSessionALPN
+    if allowHTTP t && isHTTP negotiated -- only attempt the switch for the TLS transport
+      then runHTTP socket (tlsContext h)-- ... collect data and produce values needed to run WAI Application
+      else runClient serverSignKey t h `runReaderT` env -- performs serverHandshake etc as usual
+```
+
+The web app and server live outside, so `runHttp` has to be provided by the `runSMPServer` caller.
+Additonally, Warp is using its `InternalInfo` object that's scoped to `withII` bracket.
+
+```haskell
+  runServer ini = do
+    -- ...
+
+    runWebServer ini ServerInformation {config, information} $ if sharedHttps then Nothing else webHttpsParams -- suppress serving https
+    if sharedHttps
+      then withRunHTTP staticFilesPath \attachStatic -> runSMPServer cfg (Just attachStatic) -- provide wrapped application runner
+      else runSMPServer cfg Nothing
+```
+
+### Upstream
+
+The implementation relies on a few modification to upstream code:
+- `warp-tls`: The library provides `httpOverTls`, but it wants to do handshake itself.
+  Since we have to do the handshake to switch on ALPN, the setup function has to be split.
+  This is a resonable change that may be upstreamed and nothing blocks us from using the recent version.
+- `warp`: Only the re-export of `serveConnection` is needed.
+  Unfortunately the most recent `warp` version can't be used right away due to dependency cascade around `http-5` and `auto-update-2`.
+  So a fork containing the backported re-export has to be used until the dependencies are refreshed.
+
+
+### TLS.ServerParams
+
+When a server has port sharing enabled, a new set of TLS params is loaded and combined with transport params:
+
+```haskell
+newEnv config = do
+  -- ...
+  tlsServerParams <- loadTLSServerParams caCertificateFile certificateFile privateKeyFile (alpn transportConfig)
+  sharedServerParams <- forM ((,) <$> sharedHttpsCredentials config <*> alpn transportConfig) $ \((chain, key), alpn) ->
+    let ca = Nothing -- It is possible to provide CA certificate, but it is typical for web server to use combined certificate chains
+    loadHTTPSServerParams tlsServerParams ca chain key alpn
+```
+
+`loadHTTPSServerParams` extends params with:
+1. `onALPNClientSuggest` hook gets `["h2", "http/1.1"]` added to the ALPN list which is now required.
+2. `onServerNameIndication` hook added, which upon detecting client SNI prepends the web credentials.
+3. `sharedCredentials = T.Credentials []` should be done to prevent transport credentials confusing browsers.
+    But that aborts key exchange somewhere in tls internals, so disabled for now.
+    As a workaround, another set of dummy credentials can be provided in the hope that any sane browser would reject them.
+    Like, RC4 ciphers, "impossible" digest combination, etc.
+
+### supportedParameters
+
+TLS certificate chains provided by LetsEncrypt use ECDSA/P256 and that requires extending `supportedParameters` with things disabled in transports:
+
+```haskell
+    browserCiphers =
+      [ TE.cipher_TLS13_AES128CCM8_SHA256
+      , TE.cipher_ECDHE_ECDSA_AES128CCM8_SHA256
+      , TE.cipher_ECDHE_ECDSA_AES256CCM8_SHA256
+      ]
+    browserGroups =
+      [ T.P256
+      ]
+    browserSigs =
+      [ (T.HashSHA256, T.SignatureECDSA),
+        (T.HashSHA384, T.SignatureECDSA)
+      ]
+```
+
+This may not be enough for other certificate providers.
+
+## Configuration
+
+> XXX: This is for the current implementation and should be updated.
+
+Web certificate chain is picked up from the WEB section:
+
+```ini
+[TRANSPORT]
+port: 443
+
+[WEB]
+https: 443
+cert: /etc/opt/simplex/web.cert
+key: /etc/opt/simplex/web.key
+
+# Alternatively, with a proper access configuration, the paths can point to the LE creds directly:
+# cert: /etc/letsencrypt/live/smp.hostname.tld/fullchain.pem
+# key: /etc/letsencrypt/live/smp.hostname.tld/privkey.pem
+```
+
+When `TRANSPORT.port` matches `WEB.https` the transport server becomes shared.
+
+Perhaps a more desirable option would be explicit configuration resulting in additional transported to run:
+
+```ini
+[TRANSPORT]
+port: 5223 ; pure protocol transport
+# control_port: 5224
+shared_port: 443 ; variant 1: register in TRANSPORT
+
+[WEB]
+https: 443
+cert: /etc/opt/simplex/web.cert
+key: /etc/opt/simplex/web.key
+# transport: on ; variant 2:
+```
+
+## Caveats
+
+Serving static files and the protocols togother may pose a problem for those who currently use dedicated web servers as they should switch to embedded http handlers.
+
+As before, using embedded HTTP server is increasing attack surface.
+
+Users who want to run everything on a single host will have to add and extra IP address and bind servers to specific IPs instead of 0.0.0.0.
+An amalgamated server binary can be provided that would contain both SMP and XFTP servers, where transport will dispatch connections by handshake ALPN.
+
+## Alternative: Use transports routable with reverse-proxies
+
+An "industrial" reverse proxy may do the ALPN routing, serving HTTP by itself and delegating `smp` and `xftp` to protocol servers.
+Same with the `websockets`.
+
+Since this in effect does TLS termination, the protocol servers will have to rely on credentials from protocol handshakes.

rfcs/2024-09-25-ios-notifications-2.md

@@ -0,0 +1,49 @@
+# iOS notifications delivery
+
+## Problem
+
+For iOS notifications to be delivered the client has to create credentials for notification subscription on SMP server using NKEY command and after that create a subscription on notification server using SNEW command. These two commands are sent in sequence, after the connections are created, and for it to happen the client needs to be online and in foreground.
+
+iOS users tend to close the app when it is not used, and iOS has very limited permissions for background activities, so these notification subscriptions are created with a substantial delay, and notifications do not work.
+
+This problem is distinct from and probably more common than other problems affecting notifications delivery described [here](./2024-07-06-ios-notifications.md).
+
+## Solution
+
+1. When the new connection is created, the client already knows if it needs to create notification subscription or not, based on the conversation setting (e.g., if the group is muted, the client will not create notification subscription as well.). We should extend NEW command to avoid the need to send additional NKEY command with an option to create notification subscription at the point where connection is created. NDEL would still be used to disable this notification, and NKEY will be used to re-enable it.
+
+2. In the same way we stopped using SDEL command (NDEL sends notification DELD to subscribed notification server) to delete notificaiton subscriptions from notification server, we should delegate creating notification subscription on notification server to SMP servers. Clients could use keys agreed with ntf server for e2e encryption and for command authorization to encrypt and sign instruction to create notification subscription that will be forwarded to notification server using protocol similar to SMP proxies. This will avoid the need for clients to separately contact notification servers that won't happen until they are online.
+
+3. Instead of making Ntf server trust DELD notifications, we could send deletion instructions signed by the client, which will only fail to send in case notification server is down (and they won't be sent later after server restart).
+
+Cons:
+- If SMP servers were to retain in the storage the information about which notification server is used for which queue, it would reduce metadata privacy. While currently it is not an issue, as all notification servers are known and operated by us, once there are other client apps, this can be used for app users fingerprinting, which would act as a deterrence from using new apps – but only if app users use servers of operators who are different from the app provider. To mitigate it, we could only store it in server memory and include notification instruction in subscription commands (SUB) and include notification subscription status in SUB responses. We don't need to mitigate the problem of server being able to store this information, as messaging servers can observe which notification servers connect to them anyway.
+- If SMP server is restarted before the subscription request is forwared to the notification server, then it will have to be forwarded again, once the client subscribes. The problem here is that if the client is offline, it will neither subscribe to the queue to send notification subscription request, nor receive notifications from this queue. Storing notification server and subscription request would mitigate that, as in this case we could send all pending requests on server start, without depending on client subscriptions.
+- "Small" agent will need to support connections to ntf servers and manage workers that retry sending pending subscription requests.
+- Until the client learns the public keys of notification server, it will not be able to decrypt notifications. It potentially can be mitigated by using the public key of the server returned when token is created, in this way different client keys (per-queue) will be combined with the same ntf server key (per-token).
+
+## Implementation details
+
+1. NEW and NKEY commands will need to be extended to include notification subscription request. As the notifier ID needs to be sent to notification server, this notifier ID will have to be client-generated and supplied as part of NEW command.
+
+now:
+
+```haskell
+NEW :: RcvPublicAuthKey -> RcvPublicDhKey -> Maybe BasicAuth -> SubscriptionMode -> SenderCanSecure -> Command Recipient
+NKEY :: NtfPublicAuthKey -> RcvNtfPublicDhKey -> Command Recipient
+```
+
+extended:
+
+```haskell
+NEW :: RcvPublicAuthKey -> RcvPublicDhKey -> Maybe BasicAuth -> SubscriptionMode -> SenderCanSecure -> Maybe NtfRequest -> Command Recipient
+
+data NtfRequest = NtfRequest NotifierId NtfPublicAuthKey RcvNtfPublicDhKey NtfServerRequest
+
+data NtfServerRequest = NtfServerRequest NtfServer EncSingedNtfCmd
+
+NKEY :: NtfPublicAuthKey -> RcvNtfPublicDhKey -> Maybe NtfServerRequest -> Command Recipient
+-- NotifierID is passed in entity ID field of the transmission
+```
+
+2. Notification server will need to support an additional command to receive "proxied" subscription commands, `SFWD`, that would include `NtfServerRequest`. This command can include both `SNEW` and `SDEL` commands.

rfcs/2024-11-25-journal-expiration.md

@@ -0,0 +1,15 @@
+# Expiring messages in journal storage
+
+## Problem
+
+The journal storage servers recently migrated to do not delete delivered or expired messages, they only update pointers to journal file lines. The messages are actually deleted when the whole journal file is deleted (when fully deleted or fully expired).
+
+The problem is that in case the queue stops receiving the new messages then writing of messages won't switch to the new journal file, and the current journal file containing delivered or expired messages would never be deleted.
+
+## Solution
+
+Remove current journal file and update queue_state.log during message expiration of "idle" queue (that is, without any new messages received or delivered within 3 hours) in case when:
+- the queue is "empty" after the expiration
+- the queue contains only quota marker(s), in which case move them to a new journal file and update the queue_state accordingly. Quota markers can be kept indefinitely to prevent writing the new messages to the dormant queues that reached capacity, so it's important to handle this case.
+
+Also remove current journal file when the queue is opened in case it is empty (as it would not be ever expired in case it remains empty), and also update queue_state.log

rfcs/2024-11-25-queue-blobs-2.md

@@ -0,0 +1,58 @@
+# Blob extensions for SMP queues 2 and queue storage
+
+This document evolves the design proposed [here](./2024-09-09-smp-blobs.md).
+
+## Problems
+
+In addition to problems in the first doc, we have these issues with in-memory queue record storage:
+- many queues are idle or rarely used, but they are loaded to memory, and currently just loading all queues uses 20gb RAM on each server, and takes 10 min to process, increasing downtimes during restarts.
+- adding blobs to memory would make this problem much worse.
+
+## Proposed solution
+
+Move queues to the same journalling approach as [used for messages](./2024-09-01-smp-message-storage.md) now, with independent file names in the same folders.
+
+Each queue change would be logged to its own file, and every time the queue is opened the whole file will be read and compacted to a single line - replacing one store log for all queues, with individual log files for each queue.
+
+Queue deletion would not be making a record in the file, instead it would be deleting the entire folder - it would reduce retention period for any metadata of deleted queues.
+
+We could additionally record deletions to the central log, for debugging, and reset it on every start. But in this case we should not remove folders at the point of deletion, but rather mark them as deleted and delete on restart. TBC
+
+It would also allow simplifying blob storage by having only one blob per queue - for example, limied to 16kb (a bit smaller to fit in block) for contact address queues and 4-8kb for invitations (to fit PQ keys and conversation preferences).
+
+We would also need to be able to lookup recipient ID via sender/notifier/link IDs.
+
+One possible solution is to use and load to memory a central index file. But it is likely to also consume a lot of memory and result in slow starts.
+
+Another solution that is probably better is to use the same folder structure and put notifier/sender/link files with the ID of the recipient queue inside the files. So to locate recipient queue the sender would have to locate folder containing the reference file pointing to the recipient queue and then to locate the actual queue data.
+
+## Implementation details
+
+Each queue folder would these files:
+
+- queue_state.log (and timestamped backups) - to store pointers to message journals (already implemented)
+- messages.randomBase64.log - message journals (already implemented)
+- queue_rec.log (and timestamped backups) - to log complete queue record every time it is changed (so only the last line needs to be read following the same logic as with queue_state.log, to prevent file corruption).
+- blob.data, blob.data.bak, blob.timestamp.data - files for data blobs (to make sure some copy of this file is readable/correct in case of write corruption) - the same two step overwrite process will be used as currently with store log compacting:
+  - on write: 1. if file exists, move it to .bak, 2. store new blob to .data, 3. move .bak to .timestamp.data
+  - on read: 1. if .bak exists, move it to .data 2. use .data
+
+Additional suggestion to reduce probability of queue_state.log and queue_rec.log file corruption is to do one of the following:
+- log end of lines in the beginning of the output, not in the end, to prevent the last line from being corrupted in case the previous line was not fully stored. The downside is that the file will not be EOL terminated, and there will be no confirmation that the output was fully made.
+- log EOL both in the beginning and at the end of output, and ignore empty lines in between - this would both confirm that the last line is fully logged and prevent corruption of the next line in case it was not.
+- check the last byte of the file and log EOL if it is not EOL. Probably cleanest approach, but with a small performance cost.
+
+If queue folder is a reference to the queue, it may have one of these files:
+- notifier.id
+- sender.id
+- link.id
+
+These files would contain a one line with the recipient ID of the queue. These files would never change, they can only be deleted when queue is deleted or when notifier/link is deleted.
+
+There is logic in code preventing using the same ID in different contexts, and the ID size is large enough to make any collisions unlikely (192 bits), so with correctly working code the queue folder would either have one of reference files, and nothing else, or the queue and message files from the beginning of this section. But even if the same ID is re-used in different context, it should not cause any problems as file names don't overlap.
+
+While we could store different types of references in different types of folders, it would have additional costs of maintaining 4 folder hierarchies. Instead we could use the fact that it is one hierarchy to prevent using the same ID in different contexts.
+
+## Protocol
+
+The only change in protocol is that there will be only one blob per queue, without markers (see the previous doc). Otherwise the protocol and proposed privacy improvement seem reasonable.

rfcs/done/2023-10-25-remote-control.md

@@ -53,7 +53,7 @@ The session invitation contains this data:
 - CA TLS certificate fingerprint of the controller - this is part of long term identity of the controller established during the first session, and repeated in the subsequent session announcements.
 - Session Ed25519 public key used to verify the announcement and commands - this mitigates the compromise of the long term signature key, as the controller will have to sign each command with this key first.
 - Long-term Ed25519 public key used to verify the announcement and commands - this is part of the long term controller identity.
-- Session X25519 DH key and sntrup761 KEM encapsulation key to agree session encryption (both for multicast announcement and for commands and responses in TLS), as described in https://datatracker.ietf.org/doc/draft-josefsson-ntruprime-hybrid/. The new keys are used for each session, and if client key is already available (from the previous session), the computed shared secret will be used to encrypt the announcement multicast packet. The out-of-band invitation is unencrypted. These DH public key and KEM encapsulation key are always sent unencrypted. NaCL Cryptobox is used for encryption.
+- Session X25519 DH key and sntrup761 KEM encapsulation key to agree session encryption (both for multicast announcement and for commands and responses in TLS), as described in https://datatracker.ietf.org/doc/draft-josefsson-ntruprime-hybrid/. The new keys are used for each session, and if client key is already available (from the previous session), the computed shared secret will be used to encrypt the announcement multicast packet. The out-of-band invitation is unencrypted. This DH public key is always sent unencrypted. NaCL Cryptobox is used for encryption.
 
 Host device decrypts (except the first session) and validates the invitation:
 - Session signature is valid.

scripts/main/simplex-servers-stopscript

@@ -1,30 +1,174 @@
 #!/usr/bin/env sh
 set -eu
 
-path_conf_var="/var/opt"
-path_conf_smp="$path_conf_var/simplex"
-path_conf_xftp="$path_conf_var/simplex-xftp"
-path_conf_storelog_smp="$path_conf_smp/smp-server-store.log"
-path_conf_storelog_xftp="$path_conf_xftp/file-server-store.log"
+# Common
+# ------
 date="$(date -u '+%Y-%m-%dT%H:%M:%S')"
 
-backup_smp() {
- if [ -e "$path_conf_storelog_smp" ]; then
-  cp "$path_conf_storelog_smp" "${path_conf_storelog_smp}.${date:-date-failed}"
- fi
+GRN='\033[0;32m'
+YLW='\033[1;33m'
+BLU='\033[1;34m'
+RED='\033[0;31m'
+NC='\033[0m'
+
+path_conf_var="/var/opt"
+
+smp_variables() {
+	path_conf_smp="$path_conf_var/simplex"
+	path_conf_smp_archive="$path_conf_smp/backups"
+	path_conf_smp_archive_storelog="$path_conf_smp_archive/queues"
+	path_conf_smp_archive_stats="$path_conf_smp_archive/stats"
+	path_conf_smp_archive_messages="$path_conf_smp_archive/messages"
+	path_conf_storelog_smp="$path_conf_smp/smp-server-store.log"
+	path_conf_storelog_smp_out="$path_conf_smp_archive_storelog/smp-server-store.log.${date:-date-failed}"
+
+	path_conf_stats_smp="$path_conf_smp/smp-server-stats.log"
+	path_conf_stats_smp_out="$path_conf_smp_archive_stats/smp-server-stats.log.${date:-date-failed}"
+
+	path_conf_messages_smp="$path_conf_smp/smp-server-messages.log"
+	path_conf_messages_smp_out="$path_conf_smp_archive_messages/smp-server-messages.log.${date:-date-failed}"
 }
 
-backup_xftp() {
- if [ -e "$path_conf_storelog_xftp" ]; then
-  cp "$path_conf_storelog_xftp" "${path_conf_storelog_xftp}.${date:-date-failed}"
- fi
+xftp_variables() {
+	path_conf_xftp="$path_conf_var/simplex-xftp"
+	path_conf_xftp_archive="$path_conf_xftp/backups"
+
+	path_conf_xftp_archive_storelog="$path_conf_xftp_archive/queues"
+	path_conf_xftp_archive_stats="$path_conf_xftp_archive/stats"
+
+	path_conf_storelog_xftp="$path_conf_xftp/file-server-store.log"
+	path_conf_storelog_xftp_out="$path_conf_xftp_archive_storelog/file-server-store.log.${date:-date-failed}"
+
+	path_conf_stats_xftp="$path_conf_xftp/file-server-stats.log"
+	path_conf_stats_xftp_out="$path_conf_xftp_archive_stats/file-server-stats.log.${date:-date-failed}"
 }
 
-if [ "$1" = 'smp-server' ]; then
- backup_smp
-elif [ "$1" = 'xftp-server' ]; then
- backup_xftp
-else
- backup_smp
- backup_xftp
-fi
+checks() {
+	result=${SERVICE_RESULT:-exit-code}
+	status=${EXIT_STATUS:-TERM}
+
+	case "$result" in
+		success)
+			case "$status" in
+				TERM)
+					printf "${RED}Refusing to backup files with failed service state${NC}\n"
+					exit 1
+					;;
+				*)
+					:
+					;;
+			esac
+			;;
+		*)
+			printf "${RED}Refusing to backup files with failed service state${NC}\n"
+			exit 1
+			;;
+	esac
+}
+
+smp_check() {
+	if [ ! -d "$path_conf_smp_archive_storelog" ]; then
+		mkdir -p "$path_conf_smp_archive_storelog"
+	fi
+	if [ ! -d "$path_conf_smp_archive_messages" ]; then
+		mkdir -p "$path_conf_smp_archive_messages"
+	fi
+	if [ ! -d "$path_conf_smp_archive_stats" ]; then
+		mkdir -p "$path_conf_smp_archive_stats"
+	fi
+}
+
+xftp_check() {
+	if [ ! -d "$path_conf_xftp_archive_storelog" ]; then
+		mkdir -p "$path_conf_xftp_archive_storelog"
+	fi
+	if [ ! -d "$path_conf_xftp_archive_stats" ]; then
+		mkdir -p "$path_conf_xftp_archive_stats"
+	fi
+}
+
+backup() {
+	file="$1"
+	out="$2"
+	file_type="$3"
+
+	if [ -e "$file" ]; then
+		if cp "$file" "$out"; then
+			printf "${YLW}${file_type}${NC} ${GRN}backup successful:${NC} ${BLU}%s${NC}\n" "${out}"
+		else
+			printf "${YLW}${file_type}${NC} ${RED}backup failed!${NC}\n"
+		fi
+	fi
+
+	unset file out file_type
+}
+
+cleanup() {
+	directory="$1"
+	
+	file_type="$2"
+	
+	files_date=$(find "$directory" -type f -exec stat --format="%y" {} + | awk '{print $1}' | sort -nr | uniq | awk 'NR==2')
+	
+	if [ -n "$files_date" ]; then
+		files=$(find "$directory" -type f -not -newermt "$files_date" -printf "%T@ %Tc %p\n" | sort -n | awk '{print $NF}')
+
+		if [ -n "$files" ]; then
+			printf '%s' "$files" | xargs rm -f
+			printf "${YLW}Old ${file_type} files${NC}${GRN} has been deleted:${NC}\n"
+			files_colored=$(printf '%s' "$files" | awk '{print "\033[1;34m"$0"\033[0m"}')
+			printf "${files_colored}\n"
+		fi
+	fi
+
+	unset directory file_type files_date files
+}
+
+smp_backup() {
+	backup "$path_conf_storelog_smp" "$path_conf_storelog_smp_out" 'Storelog'
+	backup "$path_conf_messages_smp" "$path_conf_messages_smp_out" 'Messages'
+	backup "$path_conf_stats_smp" "$path_conf_stats_smp_out" 'Stats'
+}
+
+smp_cleanup() {
+	cleanup "$path_conf_smp_archive_storelog" 'storelog'
+	cleanup "$path_conf_smp_archive_stats" 'stats'
+	cleanup "$path_conf_smp_archive_messages" 'messages'
+}
+
+xftp_backup() {
+	backup "$path_conf_storelog_xftp" "$path_conf_storelog_xftp_out" 'Storelog'
+	backup "$path_conf_stats_xftp" "$path_conf_stats_xftp_out" 'Stats'
+}
+
+xftp_cleanup() {
+	cleanup "$path_conf_xftp_archive_storelog" 'storelog'
+	cleanup "$path_conf_xftp_archive_stats" 'stats'
+}
+
+main() {
+	type="${1:-}"
+  
+	checks
+
+	case "$type" in
+		smp-server)
+			smp_variables
+			smp_check
+			smp_backup
+			smp_cleanup
+			;;
+		xftp-server)
+			xftp_variables
+			xftp_check
+			xftp_backup
+			xftp_cleanup
+			;;
+		*)
+			printf "${YLW}Unknown server type.${NC}\n"
+			exit 1
+			;;
+	esac 
+}
+
+main "$@"

scripts/main/simplex-servers-uninstall

@@ -13,11 +13,15 @@ fi
 printf "${RED}This action will permanently remove all configs, directories, binaries from Installation Script. Please backup any relevant configs if they are needed.${NC}\n\nPress ${GRN}ENTER${NC} to continue or ${RED}Ctrl+C${NC} to cancel installation"
 read ans
 
-systemctl stop smp-server
-systemctl stop xftp-server
+systemctl disable --now smp-server 2>/dev/null || true
+systemctl revert smp-server 2>/dev/null || true
+systemctl disable --now xftp-server 2>/dev/null || true
+systemctl revert xftp-server 2>/dev/null || true
+systemctl daemon-reload 2>/dev/null || true
 
-rm -rf /var/opt/simplex /etc/opt/simplex /var/opt/simplex-xftp /etc/opt/simplex-xftp /srv/xftp /etc/systemd/system/smp-server.service /etc/systemd/system/xftp-server.service /usr/local/bin/smp-server /usr/local/bin/xftp-server /usr/local/bin/simplex-servers-update /usr/local/bin/simplex-servers-uninstall
+rm -rf /var/opt/simplex /etc/opt/simplex /etc/opt/simplex-info /var/opt/simplex-xftp /etc/opt/simplex-xftp /srv/xftp /etc/systemd/system/smp-server.service /etc/systemd/system/xftp-server.service /usr/local/bin/smp-server /usr/local/bin/xftp-server /usr/local/bin/simplex-servers-update /usr/local/bin/simplex-servers-uninstall /usr/local/bin/simplex-servers-stopscript
 
-userdel smp && userdel xftp
+userdel smp 2>/dev/null || true
+userdel xftp 2>/dev/null || true
 
 printf "Uninstallation is complete! Thanks for trying out SimpleX!\n"

scripts/main/simplex-servers-update

@@ -2,10 +2,6 @@
 set -eu
 
 # Links to scripts/configs
-bin="https://github.com/simplex-chat/simplexmq/releases/latest/download"
-bin_smp="$bin/smp-server-ubuntu-20_04-x86-64"
-bin_xftp="$bin/xftp-server-ubuntu-20_04-x86-64"
-
 scripts="https://raw.githubusercontent.com/simplex-chat/simplexmq/stable/scripts/main"
 scripts_systemd_smp="$scripts/smp-server.service"
 scripts_systemd_xftp="$scripts/xftp-server.service"
@@ -33,6 +29,9 @@ path_tmp_bin_stopscript="$path_tmp_bin/simplex-servers-stopscript"
 path_tmp_systemd_smp="$path_tmp_bin/smp-server.service"
 path_tmp_systemd_xftp="$path_tmp_bin/xftp-server.service"
 
+path_conf_etc='/etc/opt'
+path_conf_info='/etc/opt/simplex-info'
+
 GRN='\033[0;32m'
 BLU='\033[1;36m'
 YLW='\033[1;33m'
@@ -40,8 +39,55 @@ RED='\033[0;31m'
 NC='\033[0m'
 
 # Currently, XFTP default to v0.1.0, so it doesn't make sense to check its version
-local_version="$($path_bin_smp -v | awk '{print $3}')"
-remote_version="$(curl --proto '=https' --tlsv1.2 -sSf -L https://api.github.com/repos/simplex-chat/simplexmq/releases/latest | grep -i "tag_name" | awk -F \" '{print $4}')"
+
+os_test() {
+ . /etc/os-release
+
+ case "$VERSION_ID" in
+  20.04|22.04) : ;;
+  24.04) VERSION_ID='22.04' ;;
+  *) printf "${RED}Unsupported Ubuntu version!${NC}\nPlease file Github issue with request to support Ubuntu %s: https://github.com/simplex-chat/simplexmq/issues/new\n" "$VERSION_ID" && exit 1 ;;
+ esac
+
+ version="$(printf '%s' "$VERSION_ID" | tr '.' '_')"
+ arch="$(uname -p)"
+
+ case "$arch" in
+  x86_64) arch="$(printf '%s' "$arch" | tr '_' '-')" ;;
+  *)  printf "${RED}Unsupported architecture!${NC}\nPlease file Github issue with request to support %s architecture: https://github.com/simplex-chat/simplexmq/issues/new" "$arch" && exit 1 ;;
+ esac
+
+ bin_smp="$bin/smp-server-ubuntu-${version}-${arch}"
+ bin_xftp="$bin/xftp-server-ubuntu-${version}-${arch}"
+}
+
+installed_test() {
+  set +u
+  for i in $path_conf_etc/*; do
+    if [ -d "$i" ]; then
+      case "$i" in
+        *simplex) apps="smp $apps" ;;
+        *simplex-xftp) apps="xftp $apps" ;;
+      esac
+    fi
+  done
+  set -u
+}
+
+set_version() {
+  ver="${VER:-latest}"
+
+  case "$ver" in 
+    latest)
+      bin="https://github.com/simplex-chat/simplexmq/releases/latest/download"
+      remote_version="$(curl --proto '=https' --tlsv1.2 -sSf -L https://api.github.com/repos/simplex-chat/simplexmq/releases/latest | grep -i "tag_name" | awk -F \" '{print $4}')"
+      ;;
+    *)
+      bin="https://github.com/simplex-chat/simplexmq/releases/download/${ver}"
+      remote_version="${ver}"
+      ;;
+  esac
+}
 
 update_scripts() {
   curl --proto '=https' --tlsv1.2 -sSf -L "$scripts_update" -o "$path_tmp_bin_update" && chmod +x "$path_tmp_bin_update"
@@ -56,6 +102,7 @@ update_scripts() {
     mv "$path_tmp_bin_uninstall" "$path_bin_uninstall"
     printf "${GRN}Done!${NC}\n"
   fi
+
   if diff -q "$path_bin_stopscript" "$path_tmp_bin_stopscript" > /dev/null 2>&1; then
     printf -- "- ${YLW}Stopscript script is up-to-date${NC}.\n"
     rm "$path_tmp_bin_stopscript"
@@ -64,6 +111,7 @@ update_scripts() {
     mv "$path_tmp_bin_stopscript" "$path_bin_stopscript"
     printf "${GRN}Done!${NC}\n"
   fi
+
   if diff -q "$path_bin_update" "$path_tmp_bin_update" > /dev/null 2>&1; then
     printf -- "- ${YLW}Update script is up-to-date${NC}.\n"
     rm "$path_tmp_bin_update"
@@ -71,75 +119,82 @@ update_scripts() {
     printf -- "- Updating update script..."
     mv "$path_tmp_bin_update" "$path_bin_update"
     printf "${GRN}Done!${NC}\n"
-    printf "Re-executing Update script with latest updates..."
+    printf -- "- Re-executing Update script with latest updates..."
     exec sh "$path_bin_update" "continue"
   fi
 }
 
 update_systemd() {
-  curl --proto '=https' --tlsv1.2 -sSf -L "$scripts_systemd_smp" -o "$path_tmp_systemd_smp"
-  curl --proto '=https' --tlsv1.2 -sSf -L "$scripts_systemd_xftp" -o "$path_tmp_systemd_xftp"
+  service="${1}-server"
+  eval "scripts_systemd=\$scripts_systemd_${1}"
+  eval "path_systemd=\$path_systemd_${1}"
+  eval "path_tmp_systemd=\$path_tmp_systemd_${1}"
+   
+  curl --proto '=https' --tlsv1.2 -sSf -L "$scripts_systemd" -o "$path_tmp_systemd"
 
-  if diff -q "$path_systemd_smp" "$path_tmp_systemd_smp" > /dev/null 2>&1; then
-    printf -- "- ${YLW}smp-server service is up-to-date${NC}.\n"
-    rm "$path_tmp_systemd_smp"
+  if diff -q "$path_systemd" "$path_tmp_systemd" > /dev/null 2>&1; then
+    printf -- "- ${YLW}%s service is up-to-date${NC}.\n" "$service"
+    rm "$path_tmp_systemd"
   else
-    printf -- "- Updating smp-server service..."
-    mv "$path_tmp_systemd_smp" "$path_systemd_smp"
-    systemctl daemon-reload
-    printf "${GRN}Done!${NC}\n"
-  fi
-  if diff -q "$path_systemd_xftp" "$path_tmp_systemd_xftp" > /dev/null 2>&1; then
-    printf -- "- ${YLW}xftp-server service is up-to-date${NC}.\n"
-    rm "$path_tmp_systemd_xftp"
-  else
-    printf -- "- Updating xftp-server service..."
-    mv "$path_tmp_systemd_xftp" "$path_systemd_xftp"
+    printf -- "- Updating %s service..." "$service"
+    mv "$path_tmp_systemd" "$path_systemd"
     systemctl daemon-reload
     printf "${GRN}Done!${NC}\n"
   fi
+
+  unset service scripts_systemd path_systemd path_tmp_systemd
 }
 
 update_bins() {
-  if [ "$local_version" != "$remote_version" ]; then
-    if systemctl is-active --quiet smp-server; then
-      printf -- "- Stopping smp-server service..."
-      systemctl stop smp-server
-      printf "${GRN}Done!${NC}\n"
+  service="${1}-server"
+  eval "bin=\$bin_${1}"
+  eval "path_bin=\$path_bin_${1}"
 
-      printf -- "- Updating smp-server bin to %s..." "$remote_version"
-      curl --proto '=https' --tlsv1.2 -sSf -L "$bin_smp" -o "$path_bin_smp" && chmod +x "$path_bin_smp"
-      printf "${GRN}Done!${NC}\n"
-      
-      printf -- "- Starting smp-server service..."
-      systemctl start smp-server
-      printf "${GRN}Done!${NC}\n"
-    else
-      printf -- "- Updating smp-server bin..."
-      curl --proto '=https' --tlsv1.2 -sSf -L "$bin_smp" -o "$path_bin_smp" && chmod +x "$path_bin_smp"
-      printf "${GRN}Done!${NC}\n"
+  set_ver() {
+    local_version='unset'
+    sed -i -- "s/local_version_${1}=.*/local_version_${1}='${remote_version}'/" "$path_conf_info/release"
+  }
+
+  if [ -f "$path_conf_info/release" ]; then
+    . "$path_conf_info/release" 2>/dev/null
+
+    set +u
+    eval "local_version=\$local_version_${1}"
+    set -u
+
+    if [ -z "${local_version}" ]; then
+            set_ver "$1"
     fi
+  else
+    printf 'local_version_xftp=\nlocal_version_smp=\n' > "$path_conf_info/release"
+    set_ver "$1"
+  fi
 
-    if systemctl is-active --quiet xftp-server; then
-      printf -- "- Stopping xftp-server service..."
-      systemctl stop xftp-server
+  if [ "$local_version" != "$remote_version" ]; then
+    if systemctl is-active --quiet "$service"; then
+      printf -- "- Stopping %s service..." "$service"
+      systemctl stop "$service"
       printf "${GRN}Done!${NC}\n"
 
-      printf -- "- Updating xftp-server bin to %s..." "$remote_version"
-      curl --proto '=https' --tlsv1.2 -sSf -L "$bin_xftp" -o "$path_bin_xftp" && chmod +x "$path_bin_xftp"
+      printf -- "- Updating %s to %s..." "$service" "$remote_version"
+      curl --proto '=https' --tlsv1.2 -sSf -L "$bin" -o "$path_bin" && chmod +x "$path_bin"
       printf "${GRN}Done!${NC}\n"
       
-      printf -- "- Starting xftp-server service..."
-      systemctl start xftp-server
+      printf -- "- Starting %s service..." "$service"
+      systemctl start "$service"
       printf "${GRN}Done!${NC}\n"
     else
-      printf -- "- Updating xftp-server bin..."
-      curl --proto '=https' --tlsv1.2 -sSf -L "$bin_smp" -o "$path_bin_xftp" && chmod +x "$path_bin_xftp"
+      printf -- "- Updating %s to %s..." "$service" "$remote_version"
+      curl --proto '=https' --tlsv1.2 -sSf -L "$bin" -o "$path_bin" && chmod +x "$path_bin"
       printf "${GRN}Done!${NC}\n"
     fi
   else
-    printf -- "- ${YLW}smp-server and xftp-server binaries is up-to-date${NC}.\n"
+    printf -- "- ${YLW}%s is up-to-date${NC}.\n" "$service"
   fi
+
+  set_ver "$1"
+
+  unset service bin path_bin local_version
 }
 
 checks() {
@@ -147,6 +202,12 @@ checks() {
     printf "This script is intended to be run with root privileges. Please re-run script using sudo.\n"
     exit 1
   fi
+
+  set_version
+  os_test
+  installed_test
+
+  mkdir -p $path_conf_info
 }
 
 main() {
@@ -163,10 +224,14 @@ main() {
   fi
 
   printf "Updating systemd services...\n"
-  update_systemd
+  for i in $apps; do
+    update_systemd "$i"
+  done
   
-  printf "Updating simplex server binaries...\n"
-  update_bins
+  printf "Updating simplex servers...\n"
+  for i in $apps; do
+    update_bins "$i"
+  done
 
   rm -rf "$path_tmp_bin"
 }

scripts/main/smp-server.service

@@ -9,7 +9,9 @@ ExecStart=/usr/local/bin/smp-server start +RTS -N -RTS
 ExecStopPost=/usr/local/bin/simplex-servers-stopscript smp-server
 LimitNOFILE=65535
 KillSignal=SIGINT
+TimeoutStartSec=infinity
 TimeoutStopSec=infinity
+AmbientCapabilities=CAP_NET_BIND_SERVICE
 
 [Install]
 WantedBy=multi-user.target

scripts/main/xftp-server.service

@@ -9,6 +9,7 @@ ExecStart=/usr/local/bin/xftp-server start +RTS -N -RTS
 ExecStopPost=/usr/local/bin/simplex-servers-stopscript xftp-server
 LimitNOFILE=65535
 KillSignal=SIGINT
+TimeoutStartSec=infinity
 TimeoutStopSec=infinity
 AmbientCapabilities=CAP_NET_BIND_SERVICE
    

simplexmq.cabal

@@ -1,11 +1,7 @@
 cabal-version: 1.12
 
--- This file has been generated from package.yaml by hpack version 0.35.0.
---
--- see: https://github.com/sol/hpack
-
 name:           simplexmq
-version:        5.8.2.0
+version:        6.2.1.100
 synopsis:       SimpleXMQ message broker
 description:    This package includes <./docs/Simplex-Messaging-Server.html server>,
                 <./docs/Simplex-Messaging-Client.html client> and
@@ -66,24 +62,21 @@ flag use_crypton
   manual: True
   default: True
 
+flag client_library
+  description: Don't build server-related code.
+  manual: True
+  default: False
+
 library
   exposed-modules:
       Simplex.FileTransfer.Agent
       Simplex.FileTransfer.Chunks
       Simplex.FileTransfer.Client
       Simplex.FileTransfer.Client.Agent
-      Simplex.FileTransfer.Client.Main
       Simplex.FileTransfer.Client.Presets
       Simplex.FileTransfer.Crypto
       Simplex.FileTransfer.Description
       Simplex.FileTransfer.Protocol
-      Simplex.FileTransfer.Server
-      Simplex.FileTransfer.Server.Control
-      Simplex.FileTransfer.Server.Env
-      Simplex.FileTransfer.Server.Main
-      Simplex.FileTransfer.Server.Stats
-      Simplex.FileTransfer.Server.Store
-      Simplex.FileTransfer.Server.StoreLog
       Simplex.FileTransfer.Transport
       Simplex.FileTransfer.Types
       Simplex.FileTransfer.Util
@@ -133,8 +126,10 @@ library
       Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240223_connections_wait_delivery
       Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240225_ratchet_kem
       Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240417_rcv_files_approved_relays
-      Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240518_servers_stats
       Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240624_snd_secure
+      Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240702_servers_stats
+      Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240930_ntf_tokens_to_delete
+      Simplex.Messaging.Agent.Store.SQLite.Migrations.M20241007_rcv_queues_last_broker_ts
       Simplex.Messaging.Agent.TRcvQueues
       Simplex.Messaging.Client
       Simplex.Messaging.Client.Agent
@@ -152,32 +147,12 @@ library
       Simplex.Messaging.Encoding.String
       Simplex.Messaging.Notifications.Client
       Simplex.Messaging.Notifications.Protocol
-      Simplex.Messaging.Notifications.Server
-      Simplex.Messaging.Notifications.Server.Env
-      Simplex.Messaging.Notifications.Server.Main
-      Simplex.Messaging.Notifications.Server.Push.APNS
-      Simplex.Messaging.Notifications.Server.Push.APNS.Internal
-      Simplex.Messaging.Notifications.Server.Stats
-      Simplex.Messaging.Notifications.Server.Store
-      Simplex.Messaging.Notifications.Server.StoreLog
       Simplex.Messaging.Notifications.Transport
       Simplex.Messaging.Notifications.Types
       Simplex.Messaging.Parsers
       Simplex.Messaging.Protocol
-      Simplex.Messaging.Server
-      Simplex.Messaging.Server.CLI
-      Simplex.Messaging.Server.Control
-      Simplex.Messaging.Server.Env.STM
       Simplex.Messaging.Server.Expiration
-      Simplex.Messaging.Server.Information
-      Simplex.Messaging.Server.Main
-      Simplex.Messaging.Server.MsgStore
-      Simplex.Messaging.Server.MsgStore.STM
-      Simplex.Messaging.Server.QueueStore
       Simplex.Messaging.Server.QueueStore.QueueInfo
-      Simplex.Messaging.Server.QueueStore.STM
-      Simplex.Messaging.Server.Stats
-      Simplex.Messaging.Server.StoreLog
       Simplex.Messaging.ServiceScheme
       Simplex.Messaging.Session
       Simplex.Messaging.TMap
@@ -191,7 +166,6 @@ library
       Simplex.Messaging.Transport.HTTP2.Server
       Simplex.Messaging.Transport.KeepAlive
       Simplex.Messaging.Transport.Server
-      Simplex.Messaging.Transport.WebSockets
       Simplex.Messaging.Util
       Simplex.Messaging.Version
       Simplex.Messaging.Version.Internal
@@ -200,13 +174,50 @@ library
       Simplex.RemoteControl.Discovery.Multicast
       Simplex.RemoteControl.Invitation
       Simplex.RemoteControl.Types
+  if !flag(client_library)
+      exposed-modules:
+          Simplex.FileTransfer.Client.Main
+          Simplex.FileTransfer.Server
+          Simplex.FileTransfer.Server.Control
+          Simplex.FileTransfer.Server.Env
+          Simplex.FileTransfer.Server.Main
+          Simplex.FileTransfer.Server.Stats
+          Simplex.FileTransfer.Server.Store
+          Simplex.FileTransfer.Server.StoreLog
+          Simplex.Messaging.Notifications.Server
+          Simplex.Messaging.Notifications.Server.Control
+          Simplex.Messaging.Notifications.Server.Env
+          Simplex.Messaging.Notifications.Server.Main
+          Simplex.Messaging.Notifications.Server.Push.APNS
+          Simplex.Messaging.Notifications.Server.Push.APNS.Internal
+          Simplex.Messaging.Notifications.Server.Stats
+          Simplex.Messaging.Notifications.Server.Store
+          Simplex.Messaging.Notifications.Server.StoreLog
+          Simplex.Messaging.Server
+          Simplex.Messaging.Server.CLI
+          Simplex.Messaging.Server.Control
+          Simplex.Messaging.Server.Env.STM
+          Simplex.Messaging.Server.Information
+          Simplex.Messaging.Server.Main
+          Simplex.Messaging.Server.MsgStore
+          Simplex.Messaging.Server.MsgStore.Journal
+          Simplex.Messaging.Server.MsgStore.STM
+          Simplex.Messaging.Server.MsgStore.Types
+          Simplex.Messaging.Server.NtfStore
+          Simplex.Messaging.Server.Prometheus
+          Simplex.Messaging.Server.QueueStore
+          Simplex.Messaging.Server.QueueStore.STM
+          Simplex.Messaging.Server.Stats
+          Simplex.Messaging.Server.StoreLog
+          Simplex.Messaging.Server.StoreLog.Types
+          Simplex.Messaging.Transport.WebSockets
   other-modules:
       Paths_simplexmq
   hs-source-dirs:
       src
   default-extensions:
       StrictData
-  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-unused-packages -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2
+  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-home-modules -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2
   include-dirs:
       cbits
   c-sources:
@@ -216,14 +227,12 @@ library
       crypto
   build-depends:
       aeson ==2.2.*
-    , ansi-terminal >=0.10 && <0.12
     , asn1-encoding ==0.9.*
     , asn1-types ==0.3.*
     , async ==2.2.*
     , attoparsec ==0.14.*
     , base >=4.14 && <5
     , base64-bytestring >=1.0 && <1.3
-    , case-insensitive ==1.2.*
     , composition ==1.0.*
     , constraints >=0.12 && <0.14
     , containers ==0.6.*
@@ -258,28 +267,32 @@ library
     , temporary ==1.3.*
     , time ==1.12.*
     , time-manager ==0.0.*
-    , tls >=1.7.0 && <1.8
+    , tls >=1.9.0 && <1.10
     , transformers ==0.6.*
     , unliftio ==0.2.*
     , unliftio-core ==0.2.*
-    , websockets ==0.12.*
     , yaml ==0.11.*
     , zstd ==0.1.3.*
   default-language: Haskell2010
   if flag(swift)
     cpp-options: -DswiftJSON
+  if !flag(client_library)
+    build-depends:
+        case-insensitive ==1.2.*
+      , hashable ==1.4.*
+      , websockets ==0.12.*
   if impl(ghc >= 9.6.2)
     build-depends:
         bytestring ==0.11.*
-      , template-haskell ==2.20.*
       , text >=2.0.1 && <2.2
   if impl(ghc < 9.6.2)
     build-depends:
         bytestring ==0.10.*
-      , template-haskell ==2.16.*
       , text >=1.2.3.0 && <1.3
 
 executable ntf-server
+  if flag(client_library)
+      buildable: False
   main-is: Main.hs
   other-modules:
       Paths_simplexmq
@@ -287,74 +300,16 @@ executable ntf-server
       apps/ntf-server
   default-extensions:
       StrictData
-  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-unused-packages -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2 -threaded -rtsopts
+  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2 -threaded -rtsopts
   build-depends:
-      aeson ==2.2.*
-    , ansi-terminal >=0.10 && <0.12
-    , asn1-encoding ==0.9.*
-    , asn1-types ==0.3.*
-    , async ==2.2.*
-    , attoparsec ==0.14.*
-    , base >=4.14 && <5
-    , base64-bytestring >=1.0 && <1.3
-    , case-insensitive ==1.2.*
-    , composition ==1.0.*
-    , constraints >=0.12 && <0.14
-    , containers ==0.6.*
-    , crypton ==0.34.*
-    , crypton-x509 ==1.7.*
-    , crypton-x509-store ==1.6.*
-    , crypton-x509-validation ==1.6.*
-    , cryptostore ==0.3.*
-    , data-default ==0.7.*
-    , direct-sqlcipher ==2.3.*
-    , directory ==1.3.*
-    , filepath ==1.4.*
-    , hourglass ==0.2.*
-    , http-types ==0.12.*
-    , http2 >=4.2.2 && <4.3
-    , ini ==0.4.1
-    , iproute ==1.7.*
-    , iso8601-time ==0.1.*
-    , memory ==0.18.*
-    , mtl >=2.3.1 && <3.0
-    , network >=3.1.2.7 && <3.2
-    , network-info ==0.2.*
-    , network-transport ==0.5.6
-    , network-udp ==0.0.*
-    , optparse-applicative >=0.15 && <0.17
-    , process ==1.6.*
-    , random >=1.1 && <1.3
-    , simple-logger ==0.1.*
+      base
+    , simple-logger
     , simplexmq
-    , socks ==0.6.*
-    , sqlcipher-simple ==0.4.*
-    , stm ==2.5.*
-    , temporary ==1.3.*
-    , time ==1.12.*
-    , time-manager ==0.0.*
-    , tls >=1.7.0 && <1.8
-    , transformers ==0.6.*
-    , unliftio ==0.2.*
-    , unliftio-core ==0.2.*
-    , websockets ==0.12.*
-    , yaml ==0.11.*
-    , zstd ==0.1.3.*
   default-language: Haskell2010
-  if flag(swift)
-    cpp-options: -DswiftJSON
-  if impl(ghc >= 9.6.2)
-    build-depends:
-        bytestring ==0.11.*
-      , template-haskell ==2.20.*
-      , text >=2.0.1 && <2.2
-  if impl(ghc < 9.6.2)
-    build-depends:
-        bytestring ==0.10.*
-      , template-haskell ==2.16.*
-      , text >=1.2.3.0 && <1.3
 
 executable smp-server
+  if flag(client_library)
+      buildable: False
   main-is: Main.hs
   other-modules:
       Static
@@ -365,78 +320,27 @@ executable smp-server
       apps/smp-server/web
   default-extensions:
       StrictData
-  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-unused-packages -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2 -threaded -rtsopts
+  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2 -threaded -rtsopts
   build-depends:
-      aeson ==2.2.*
-    , ansi-terminal >=0.10 && <0.12
-    , asn1-encoding ==0.9.*
-    , asn1-types ==0.3.*
-    , async ==2.2.*
-    , attoparsec ==0.14.*
-    , base >=4.14 && <5
-    , base64-bytestring >=1.0 && <1.3
-    , case-insensitive ==1.2.*
-    , composition ==1.0.*
-    , constraints >=0.12 && <0.14
-    , containers ==0.6.*
-    , crypton ==0.34.*
-    , crypton-x509 ==1.7.*
-    , crypton-x509-store ==1.6.*
-    , crypton-x509-validation ==1.6.*
-    , cryptostore ==0.3.*
-    , data-default ==0.7.*
-    , direct-sqlcipher ==2.3.*
-    , directory ==1.3.*
+      base
+    , bytestring
+    , directory
     , file-embed
-    , filepath ==1.4.*
-    , hourglass ==0.2.*
-    , http-types ==0.12.*
-    , http2 >=4.2.2 && <4.3
-    , ini ==0.4.1
-    , iproute ==1.7.*
-    , iso8601-time ==0.1.*
-    , memory ==0.18.*
-    , mtl >=2.3.1 && <3.0
-    , network >=3.1.2.7 && <3.2
-    , network-info ==0.2.*
-    , network-transport ==0.5.6
-    , network-udp ==0.0.*
-    , optparse-applicative >=0.15 && <0.17
-    , process ==1.6.*
-    , random >=1.1 && <1.3
-    , simple-logger ==0.1.*
+    , filepath
+    , network
+    , simple-logger
     , simplexmq
-    , socks ==0.6.*
-    , sqlcipher-simple ==0.4.*
-    , stm ==2.5.*
-    , temporary ==1.3.*
-    , time ==1.12.*
-    , time-manager ==0.0.*
-    , tls >=1.7.0 && <1.8
-    , transformers ==0.6.*
-    , unliftio ==0.2.*
-    , unliftio-core ==0.2.*
+    , text
+    , unliftio
+    , wai
     , wai-app-static
-    , warp
-    , warp-tls
-    , websockets ==0.12.*
-    , yaml ==0.11.*
-    , zstd ==0.1.3.*
+    , warp ==3.3.30
+    , warp-tls ==3.4.7
   default-language: Haskell2010
-  if flag(swift)
-    cpp-options: -DswiftJSON
-  if impl(ghc >= 9.6.2)
-    build-depends:
-        bytestring ==0.11.*
-      , template-haskell ==2.20.*
-      , text >=2.0.1 && <2.2
-  if impl(ghc < 9.6.2)
-    build-depends:
-        bytestring ==0.10.*
-      , template-haskell ==2.16.*
-      , text >=1.2.3.0 && <1.3
 
 executable xftp
+  if flag(client_library)
+      buildable: False
   main-is: Main.hs
   other-modules:
       Paths_simplexmq
@@ -444,74 +348,15 @@ executable xftp
       apps/xftp
   default-extensions:
       StrictData
-  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-unused-packages -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2 -threaded -rtsopts
+  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2 -threaded -rtsopts
   build-depends:
-      aeson ==2.2.*
-    , ansi-terminal >=0.10 && <0.12
-    , asn1-encoding ==0.9.*
-    , asn1-types ==0.3.*
-    , async ==2.2.*
-    , attoparsec ==0.14.*
-    , base >=4.14 && <5
-    , base64-bytestring >=1.0 && <1.3
-    , case-insensitive ==1.2.*
-    , composition ==1.0.*
-    , constraints >=0.12 && <0.14
-    , containers ==0.6.*
-    , crypton ==0.34.*
-    , crypton-x509 ==1.7.*
-    , crypton-x509-store ==1.6.*
-    , crypton-x509-validation ==1.6.*
-    , cryptostore ==0.3.*
-    , data-default ==0.7.*
-    , direct-sqlcipher ==2.3.*
-    , directory ==1.3.*
-    , filepath ==1.4.*
-    , hourglass ==0.2.*
-    , http-types ==0.12.*
-    , http2 >=4.2.2 && <4.3
-    , ini ==0.4.1
-    , iproute ==1.7.*
-    , iso8601-time ==0.1.*
-    , memory ==0.18.*
-    , mtl >=2.3.1 && <3.0
-    , network >=3.1.2.7 && <3.2
-    , network-info ==0.2.*
-    , network-transport ==0.5.6
-    , network-udp ==0.0.*
-    , optparse-applicative >=0.15 && <0.17
-    , process ==1.6.*
-    , random >=1.1 && <1.3
-    , simple-logger ==0.1.*
+      base
     , simplexmq
-    , socks ==0.6.*
-    , sqlcipher-simple ==0.4.*
-    , stm ==2.5.*
-    , temporary ==1.3.*
-    , time ==1.12.*
-    , time-manager ==0.0.*
-    , tls >=1.7.0 && <1.8
-    , transformers ==0.6.*
-    , unliftio ==0.2.*
-    , unliftio-core ==0.2.*
-    , websockets ==0.12.*
-    , yaml ==0.11.*
-    , zstd ==0.1.3.*
   default-language: Haskell2010
-  if flag(swift)
-    cpp-options: -DswiftJSON
-  if impl(ghc >= 9.6.2)
-    build-depends:
-        bytestring ==0.11.*
-      , template-haskell ==2.20.*
-      , text >=2.0.1 && <2.2
-  if impl(ghc < 9.6.2)
-    build-depends:
-        bytestring ==0.10.*
-      , template-haskell ==2.16.*
-      , text >=1.2.3.0 && <1.3
 
 executable xftp-server
+  if flag(client_library)
+      buildable: False
   main-is: Main.hs
   other-modules:
       Paths_simplexmq
@@ -519,74 +364,16 @@ executable xftp-server
       apps/xftp-server
   default-extensions:
       StrictData
-  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-unused-packages -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2 -threaded -rtsopts
+  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2 -threaded -rtsopts
   build-depends:
-      aeson ==2.2.*
-    , ansi-terminal >=0.10 && <0.12
-    , asn1-encoding ==0.9.*
-    , asn1-types ==0.3.*
-    , async ==2.2.*
-    , attoparsec ==0.14.*
-    , base >=4.14 && <5
-    , base64-bytestring >=1.0 && <1.3
-    , case-insensitive ==1.2.*
-    , composition ==1.0.*
-    , constraints >=0.12 && <0.14
-    , containers ==0.6.*
-    , crypton ==0.34.*
-    , crypton-x509 ==1.7.*
-    , crypton-x509-store ==1.6.*
-    , crypton-x509-validation ==1.6.*
-    , cryptostore ==0.3.*
-    , data-default ==0.7.*
-    , direct-sqlcipher ==2.3.*
-    , directory ==1.3.*
-    , filepath ==1.4.*
-    , hourglass ==0.2.*
-    , http-types ==0.12.*
-    , http2 >=4.2.2 && <4.3
-    , ini ==0.4.1
-    , iproute ==1.7.*
-    , iso8601-time ==0.1.*
-    , memory ==0.18.*
-    , mtl >=2.3.1 && <3.0
-    , network >=3.1.2.7 && <3.2
-    , network-info ==0.2.*
-    , network-transport ==0.5.6
-    , network-udp ==0.0.*
-    , optparse-applicative >=0.15 && <0.17
-    , process ==1.6.*
-    , random >=1.1 && <1.3
-    , simple-logger ==0.1.*
+      base
+    , simple-logger
     , simplexmq
-    , socks ==0.6.*
-    , sqlcipher-simple ==0.4.*
-    , stm ==2.5.*
-    , temporary ==1.3.*
-    , time ==1.12.*
-    , time-manager ==0.0.*
-    , tls >=1.7.0 && <1.8
-    , transformers ==0.6.*
-    , unliftio ==0.2.*
-    , unliftio-core ==0.2.*
-    , websockets ==0.12.*
-    , yaml ==0.11.*
-    , zstd ==0.1.3.*
   default-language: Haskell2010
-  if flag(swift)
-    cpp-options: -DswiftJSON
-  if impl(ghc >= 9.6.2)
-    build-depends:
-        bytestring ==0.11.*
-      , template-haskell ==2.20.*
-      , text >=2.0.1 && <2.2
-  if impl(ghc < 9.6.2)
-    build-depends:
-        bytestring ==0.10.*
-      , template-haskell ==2.16.*
-      , text >=1.2.3.0 && <1.3
 
 test-suite simplexmq-test
+  if flag(client_library)
+      buildable: False
   type: exitcode-stdio-1.0
   main-is: Test.hs
   other-modules:
@@ -598,13 +385,17 @@ test-suite simplexmq-test
       AgentTests.MigrationTests
       AgentTests.NotificationTests
       AgentTests.SchemaDump
+      AgentTests.ServerChoice
       AgentTests.SQLiteTests
       CLITests
       CoreTests.BatchingTests
       CoreTests.CryptoFileTests
       CoreTests.CryptoTests
       CoreTests.EncodingTests
+      CoreTests.MsgStoreTests
       CoreTests.RetryIntervalTests
+      CoreTests.SOCKSSettings
+      CoreTests.StoreLogTests
       CoreTests.TRcvQueuesTests
       CoreTests.UtilTests
       CoreTests.VersionRangeTests
@@ -621,83 +412,61 @@ test-suite simplexmq-test
       XFTPCLI
       XFTPClient
       XFTPServerTests
+      Static
+      Static.Embedded
       Paths_simplexmq
   hs-source-dirs:
       tests
+      apps/smp-server/web
   default-extensions:
       StrictData
-  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-unused-packages -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2 -threaded -rtsopts -with-rtsopts=-A64M -with-rtsopts=-N1
+  ghc-options: -Weverything -Wno-missing-exported-signatures -Wno-missing-import-lists -Wno-missed-specialisations -Wno-all-missed-specialisations -Wno-unsafe -Wno-safe -Wno-missing-local-signatures -Wno-missing-kind-signatures -Wno-missing-deriving-strategies -Wno-monomorphism-restriction -Wno-prepositive-qualified-module -Wno-implicit-prelude -Wno-missing-safe-haskell-mode -Wno-missing-export-lists -Wno-partial-fields -Wcompat -Werror=incomplete-record-updates -Werror=incomplete-patterns -Werror=incomplete-uni-patterns -Werror=missing-methods -Werror=tabs -Wredundant-constraints -Wincomplete-record-updates -Wunused-type-patterns -O2 -threaded -rtsopts -with-rtsopts=-A64M -with-rtsopts=-N1
   build-depends:
-      HUnit ==1.6.*
-    , QuickCheck ==2.14.*
-    , aeson ==2.2.*
-    , ansi-terminal >=0.10 && <0.12
-    , asn1-encoding ==0.9.*
-    , asn1-types ==0.3.*
-    , async ==2.2.*
-    , attoparsec ==0.14.*
-    , base >=4.14 && <5
-    , base64-bytestring >=1.0 && <1.3
-    , case-insensitive ==1.2.*
-    , composition ==1.0.*
-    , constraints >=0.12 && <0.14
-    , containers ==0.6.*
-    , crypton ==0.34.*
-    , crypton-x509 ==1.7.*
-    , crypton-x509-store ==1.6.*
-    , crypton-x509-validation ==1.6.*
-    , cryptostore ==0.3.*
-    , data-default ==0.7.*
+      base
+    , aeson
+    , async
+    , base64-bytestring
+    , bytestring
+    , containers
+    , crypton
+    , crypton-x509
+    , crypton-x509-store
+    , crypton-x509-validation
     , deepseq ==1.4.*
-    , direct-sqlcipher ==2.3.*
-    , directory ==1.3.*
-    , filepath ==1.4.*
+    , directory
+    , file-embed
+    , filepath
     , generic-random ==1.5.*
-    , hourglass ==0.2.*
+    , hashable
     , hspec ==2.11.*
-    , hspec-core ==2.11.*
-    , http-types ==0.12.*
-    , http2 >=4.2.2 && <4.3
-    , ini ==0.4.1
-    , iproute ==1.7.*
-    , iso8601-time ==0.1.*
+    , http-client
+    , http-types
+    , http2
+    , HUnit ==1.6.*
+    , ini
+    , iso8601-time
     , main-tester ==0.2.*
-    , memory ==0.18.*
-    , mtl >=2.3.1 && <3.0
-    , network >=3.1.2.7 && <3.2
-    , network-info ==0.2.*
-    , network-transport ==0.5.6
-    , network-udp ==0.0.*
-    , optparse-applicative >=0.15 && <0.17
-    , process ==1.6.*
-    , random >=1.1 && <1.3
+    , memory
+    , mtl
+    , network
+    , process
+    , QuickCheck ==2.14.*
+    , random
     , silently ==1.2.*
-    , simple-logger ==0.1.*
+    , simple-logger
     , simplexmq
-    , socks ==0.6.*
-    , sqlcipher-simple ==0.4.*
-    , stm ==2.5.*
-    , temporary ==1.3.*
-    , time ==1.12.*
-    , time-manager ==0.0.*
+    , sqlcipher-simple
+    , stm
+    , text
+    , time
     , timeit ==2.0.*
-    , tls >=1.7.0 && <1.8
-    , transformers ==0.6.*
-    , unliftio ==0.2.*
-    , unliftio-core ==0.2.*
-    , websockets ==0.12.*
-    , yaml ==0.11.*
-    , zstd ==0.1.3.*
+    , transformers
+    , unliftio
+    , unliftio-core
+    , unordered-containers
+    , wai
+    , wai-app-static
+    , warp
+    , warp-tls
+    , yaml
   default-language: Haskell2010
-  if flag(swift)
-    cpp-options: -DswiftJSON
-  if impl(ghc >= 9.6.2)
-    build-depends:
-        bytestring ==0.11.*
-      , template-haskell ==2.20.*
-      , text >=2.0.1 && <2.2
-  if impl(ghc < 9.6.2)
-    build-depends:
-        bytestring ==0.10.*
-      , template-haskell ==2.16.*
-      , text >=1.2.3.0 && <1.3

src/Simplex/FileTransfer/Agent.hs

@@ -12,6 +12,7 @@
 
 module Simplex.FileTransfer.Agent
   ( startXFTPWorkers,
+    startXFTPSndWorkers,
     closeXFTPAgent,
     toFSFilePath,
     -- Receiving files
@@ -42,15 +43,15 @@ import Data.Either (partitionEithers, rights)
 import Data.Int (Int64)
 import Data.List (foldl', partition, sortOn)
 import qualified Data.List.NonEmpty as L
-import Data.Map (Map)
+import Data.Map.Strict (Map)
 import qualified Data.Map.Strict as M
-import Data.Maybe (mapMaybe)
+import Data.Maybe (fromMaybe, mapMaybe)
 import qualified Data.Set as S
 import Data.Text (Text)
 import Data.Time.Clock (getCurrentTime)
 import Data.Time.Format (defaultTimeLocale, formatTime)
-import Simplex.FileTransfer.Client (XFTPChunkSpec (..))
-import Simplex.FileTransfer.Client.Main
+import Simplex.FileTransfer.Chunks (toKB)
+import Simplex.FileTransfer.Client (XFTPChunkSpec (..), getChunkDigest, prepareChunkSizes, prepareChunkSpecs, singleChunkSize)
 import Simplex.FileTransfer.Crypto
 import Simplex.FileTransfer.Description
 import Simplex.FileTransfer.Protocol (FileParty (..), SFileParty (..))
@@ -72,7 +73,7 @@ import qualified Simplex.Messaging.Crypto.File as CF
 import qualified Simplex.Messaging.Crypto.Lazy as LC
 import Simplex.Messaging.Encoding
 import Simplex.Messaging.Encoding.String (strDecode, strEncode)
-import Simplex.Messaging.Protocol (EntityId, ProtocolServer, ProtocolType (..), XFTPServer)
+import Simplex.Messaging.Protocol (ProtocolServer, ProtocolType (..), XFTPServer)
 import qualified Simplex.Messaging.TMap as TM
 import Simplex.Messaging.Util (catchAll_, liftError, tshow, unlessM, whenM)
 import System.FilePath (takeFileName, (</>))
@@ -81,13 +82,21 @@ import UnliftIO.Directory
 import qualified UnliftIO.Exception as E
 
 startXFTPWorkers :: AgentClient -> Maybe FilePath -> AM ()
-startXFTPWorkers c workDir = do
+startXFTPWorkers = startXFTPWorkers_ True
+{-# INLINE startXFTPWorkers #-}
+
+startXFTPSndWorkers :: AgentClient -> Maybe FilePath -> AM ()
+startXFTPSndWorkers = startXFTPWorkers_ False
+{-# INLINE startXFTPSndWorkers #-}
+
+startXFTPWorkers_ :: Bool -> AgentClient -> Maybe FilePath -> AM ()
+startXFTPWorkers_ allWorkers c workDir = do
   wd <- asks $ xftpWorkDir . xftpAgent
   atomically $ writeTVar wd workDir
   cfg <- asks config
-  startRcvFiles cfg
+  when allWorkers $ startRcvFiles cfg
   startSndFiles cfg
-  startDelFiles cfg
+  when allWorkers $ startDelFiles cfg
   where
     startRcvFiles :: AgentConfig -> AM ()
     startRcvFiles AgentConfig {rcvFilesTTL} = do
@@ -174,16 +183,18 @@ runXFTPRcvWorker c srv Worker {doWork} = do
   cfg <- asks config
   forever $ do
     lift $ waitForWork doWork
-    atomically $ assertAgentForeground c
+    liftIO $ assertAgentForeground c
     runXFTPOperation cfg
   where
     runXFTPOperation :: AgentConfig -> AM ()
     runXFTPOperation AgentConfig {rcvFilesTTL, reconnectInterval = ri, xftpConsecutiveRetries} =
       withWork c doWork (\db -> getNextRcvChunkToDownload db srv rcvFilesTTL) $ \case
-        (RcvFileChunk {rcvFileId, rcvFileEntityId, fileTmpPath, replicas = []}, _) -> rcvWorkerInternalError c rcvFileId rcvFileEntityId (Just fileTmpPath) (INTERNAL "chunk has no replicas")
-        (fc@RcvFileChunk {userId, rcvFileId, rcvFileEntityId, digest, fileTmpPath, replicas = replica@RcvFileChunkReplica {rcvChunkReplicaId, server, delay} : _}, approvedRelays) -> do
+        (RcvFileChunk {rcvFileId, rcvFileEntityId, fileTmpPath, replicas = []}, _, redirectEntityId_) ->
+          rcvWorkerInternalError c rcvFileId rcvFileEntityId redirectEntityId_ (Just fileTmpPath) (INTERNAL "chunk has no replicas")
+        (fc@RcvFileChunk {userId, rcvFileId, rcvFileEntityId, digest, fileTmpPath, replicas = replica@RcvFileChunkReplica {rcvChunkReplicaId, server, delay} : _}, approvedRelays, redirectEntityId_) -> do
           let ri' = maybe ri (\d -> ri {initialInterval = d, increaseAfter = 0}) delay
           withRetryIntervalLimit xftpConsecutiveRetries ri' $ \delay' loop -> do
+            liftIO $ waitWhileSuspended c
             liftIO $ waitForUserNetwork c
             atomically $ incXFTPServerStat c userId srv downloadAttempts
             downloadFileChunk fc replica approvedRelays
@@ -191,25 +202,26 @@ runXFTPRcvWorker c srv Worker {doWork} = do
           where
             retryLoop loop e replicaDelay = do
               flip catchAgentError (\_ -> pure ()) $ do
-                when (serverHostError e) $ notify c rcvFileEntityId $ RFWARN e
+                when (serverHostError e) $ notify c (fromMaybe rcvFileEntityId redirectEntityId_) (RFWARN e)
                 liftIO $ closeXFTPServerClient c userId server digest
                 withStore' c $ \db -> updateRcvChunkReplicaDelay db rcvChunkReplicaId replicaDelay
-              atomically $ assertAgentForeground c
+              liftIO $ assertAgentForeground c
               loop
             retryDone e = do
               atomically . incXFTPServerStat c userId srv $ case e of
                 XFTP _ XFTP.AUTH -> downloadAuthErrs
                 _ -> downloadErrs
-              rcvWorkerInternalError c rcvFileId rcvFileEntityId (Just fileTmpPath) e
+              rcvWorkerInternalError c rcvFileId rcvFileEntityId redirectEntityId_ (Just fileTmpPath) e
     downloadFileChunk :: RcvFileChunk -> RcvFileChunkReplica -> Bool -> AM ()
     downloadFileChunk RcvFileChunk {userId, rcvFileId, rcvFileEntityId, rcvChunkId, chunkNo, chunkSize, digest, fileTmpPath} replica approvedRelays = do
       unlessM ((approvedRelays ||) <$> ipAddressProtected') $ throwE $ FILE NOT_APPROVED
       fsFileTmpPath <- lift $ toFSFilePath fileTmpPath
       chunkPath <- uniqueCombine fsFileTmpPath $ show chunkNo
-      let chunkSpec = XFTPRcvChunkSpec chunkPath (unFileSize chunkSize) (unFileDigest digest)
+      let chSize = unFileSize chunkSize
+          chunkSpec = XFTPRcvChunkSpec chunkPath chSize (unFileDigest digest)
           relChunkPath = fileTmpPath </> takeFileName chunkPath
       agentXFTPDownloadChunk c userId digest replica chunkSpec
-      atomically $ waitUntilForeground c
+      liftIO $ waitUntilForeground c
       (entityId, complete, progress) <- withStore c $ \db -> runExceptT $ do
         liftIO $ updateRcvFileChunkReceived db (rcvChunkReplicaId replica) rcvChunkId relChunkPath
         RcvFile {size = FileSize currentSize, chunks, redirect} <- ExceptT $ getRcvFile db rcvFileId
@@ -221,13 +233,14 @@ runXFTPRcvWorker c srv Worker {doWork} = do
         liftIO . when complete $ updateRcvFileStatus db rcvFileId RFSReceived
         pure (entityId, complete, RFPROG rcvd total)
       atomically $ incXFTPServerStat c userId srv downloads
+      atomically $ incXFTPServerSizeStat c userId srv downloadsSize (fromIntegral $ toKB chSize)
       notify c entityId progress
       when complete . lift . void $
         getXFTPRcvWorker True c Nothing
       where
         ipAddressProtected' :: AM Bool
         ipAddressProtected' = do
-          cfg <- liftIO $ getNetworkConfig' c
+          cfg <- liftIO $ getFastNetworkConfig c
           pure $ ipAddressProtected cfg srv
         receivedSize :: [RcvFileChunk] -> Int64
         receivedSize = foldl' (\sz ch -> sz + receivedChunkSize ch) 0
@@ -249,25 +262,25 @@ retryOnError name loop done e = do
     then loop
     else done
 
-rcvWorkerInternalError :: AgentClient -> DBRcvFileId -> RcvFileId -> Maybe FilePath -> AgentErrorType -> AM ()
-rcvWorkerInternalError c rcvFileId rcvFileEntityId tmpPath err = do
+rcvWorkerInternalError :: AgentClient -> DBRcvFileId -> RcvFileId -> Maybe RcvFileId -> Maybe FilePath -> AgentErrorType -> AM ()
+rcvWorkerInternalError c rcvFileId rcvFileEntityId redirectEntityId_ tmpPath err = do
   lift $ forM_ tmpPath (removePath <=< toFSFilePath)
   withStore' c $ \db -> updateRcvFileError db rcvFileId (show err)
-  notify c rcvFileEntityId $ RFERR err
+  notify c (fromMaybe rcvFileEntityId redirectEntityId_) (RFERR err)
 
 runXFTPRcvLocalWorker :: AgentClient -> Worker -> AM ()
 runXFTPRcvLocalWorker c Worker {doWork} = do
   cfg <- asks config
   forever $ do
     lift $ waitForWork doWork
-    atomically $ assertAgentForeground c
+    liftIO $ assertAgentForeground c
     runXFTPOperation cfg
   where
     runXFTPOperation :: AgentConfig -> AM ()
     runXFTPOperation AgentConfig {rcvFilesTTL} =
       withWork c doWork (`getNextRcvFileToDecrypt` rcvFilesTTL) $
-        \f@RcvFile {rcvFileId, rcvFileEntityId, tmpPath} ->
-          decryptFile f `catchAgentError` rcvWorkerInternalError c rcvFileId rcvFileEntityId tmpPath
+        \f@RcvFile {rcvFileId, rcvFileEntityId, tmpPath, redirect} ->
+          decryptFile f `catchAgentError` rcvWorkerInternalError c rcvFileId rcvFileEntityId (redirectEntityId <$> redirect) tmpPath
     decryptFile :: RcvFile -> AM ()
     decryptFile RcvFile {rcvFileId, rcvFileEntityId, size, digest, key, nonce, tmpPath, saveFile, status, chunks, redirect} = do
       let CryptoFile savePath cfArgs = saveFile
@@ -286,12 +299,12 @@ runXFTPRcvLocalWorker c Worker {doWork} = do
         Nothing -> do
           notify c rcvFileEntityId $ RFDONE fsSavePath
           lift $ forM_ tmpPath (removePath <=< toFSFilePath)
-          atomically $ waitUntilForeground c
+          liftIO $ waitUntilForeground c
           withStore' c (`updateRcvFileComplete` rcvFileId)
         Just RcvFileRedirect {redirectFileInfo, redirectDbId} -> do
           let RedirectFileInfo {size = redirectSize, digest = redirectDigest} = redirectFileInfo
           lift $ forM_ tmpPath (removePath <=< toFSFilePath)
-          atomically $ waitUntilForeground c
+          liftIO $ waitUntilForeground c
           withStore' c (`updateRcvFileComplete` rcvFileId)
           -- proceed with redirect
           yaml <- liftError (FILE . FILE_IO . show) (CF.readFile $ CryptoFile fsSavePath cfArgs) `agentFinally` (lift $ toFSFilePath fsSavePath >>= removePath)
@@ -333,7 +346,7 @@ xftpDeleteRcvFiles' c rcvFileEntityIds = do
     batchFiles :: (DB.Connection -> DBRcvFileId -> IO a) -> [RcvFile] -> AM' [Either AgentErrorType a]
     batchFiles f rcvFiles = withStoreBatch' c $ \db -> map (\RcvFile {rcvFileId} -> f db rcvFileId) rcvFiles
 
-notify :: forall m e. (MonadIO m, AEntityI e) => AgentClient -> EntityId -> AEvent e -> m ()
+notify :: forall m e. (MonadIO m, AEntityI e) => AgentClient -> AEntityId -> AEvent e -> m ()
 notify c entId cmd = atomically $ writeTBQueue (subQ c) ("", entId, AEvt (sAEntity @e) cmd)
 
 xftpSendFile' :: AgentClient -> UserId -> CryptoFile -> Int -> AM SndFileId
@@ -379,7 +392,7 @@ runXFTPSndPrepareWorker c Worker {doWork} = do
   cfg <- asks config
   forever $ do
     lift $ waitForWork doWork
-    atomically $ assertAgentForeground c
+    liftIO $ assertAgentForeground c
     runXFTPOperation cfg
   where
     runXFTPOperation :: AgentConfig -> AM ()
@@ -441,31 +454,32 @@ runXFTPSndPrepareWorker c Worker {doWork} = do
           SndFileChunkReplica {server} : _ -> Right server
         createChunk :: Int -> SndFileChunk -> AM (ProtocolServer 'PXFTP)
         createChunk numRecipients' ch = do
-          atomically $ assertAgentForeground c
+          liftIO $ assertAgentForeground c
           (replica, ProtoServerWithAuth srv _) <- tryCreate
           withStore' c $ \db -> createSndFileReplica db ch replica
           pure srv
           where
             tryCreate = do
-              usedSrvs <- newTVarIO ([] :: [XFTPServer])
+              triedHosts <- newTVarIO S.empty
               let AgentClient {xftpServers} = c
-              userSrvCount <- length <$> atomically (TM.lookup userId xftpServers)
+              userSrvCount <- liftIO $ length <$> TM.lookupIO userId xftpServers
               withRetryIntervalCount (riFast ri) $ \n _ loop -> do
+                liftIO $ waitWhileSuspended c
                 liftIO $ waitForUserNetwork c
                 let triedAllSrvs = n > userSrvCount
-                createWithNextSrv usedSrvs
+                createWithNextSrv triedHosts
                   `catchAgentError` \e -> retryOnError "XFTP prepare worker" (retryLoop loop triedAllSrvs e) (throwE e) e
               where
                 -- we don't do closeXFTPServerClient here to not risk closing connection for concurrent chunk upload
                 retryLoop loop triedAllSrvs e = do
                   flip catchAgentError (\_ -> pure ()) $ do
                     when (triedAllSrvs && serverHostError e) $ notify c sndFileEntityId $ SFWARN e
-                  atomically $ assertAgentForeground c
+                  liftIO $ assertAgentForeground c
                   loop
-            createWithNextSrv usedSrvs = do
+            createWithNextSrv triedHosts = do
               deleted <- withStore' c $ \db -> getSndFileDeleted db sndFileId
               when deleted $ throwE $ FILE NO_FILE
-              withNextSrv c userId usedSrvs [] $ \srvAuth -> do
+              withNextSrv c userId storageSrvs triedHosts [] $ \srvAuth -> do
                 replica <- agentXFTPNewChunk c ch numRecipients' srvAuth
                 pure (replica, srvAuth)
 
@@ -480,7 +494,7 @@ runXFTPSndWorker c srv Worker {doWork} = do
   cfg <- asks config
   forever $ do
     lift $ waitForWork doWork
-    atomically $ assertAgentForeground c
+    liftIO $ assertAgentForeground c
     runXFTPOperation cfg
   where
     runXFTPOperation :: AgentConfig -> AM ()
@@ -490,6 +504,7 @@ runXFTPSndWorker c srv Worker {doWork} = do
         fc@SndFileChunk {userId, sndFileId, sndFileEntityId, filePrefixPath, digest, replicas = replica@SndFileChunkReplica {sndChunkReplicaId, server, delay} : _} -> do
           let ri' = maybe ri (\d -> ri {initialInterval = d, increaseAfter = 0}) delay
           withRetryIntervalLimit xftpConsecutiveRetries ri' $ \delay' loop -> do
+            liftIO $ waitWhileSuspended c
             liftIO $ waitForUserNetwork c
             atomically $ incXFTPServerStat c userId srv uploadAttempts
             uploadFileChunk cfg fc replica
@@ -500,20 +515,20 @@ runXFTPSndWorker c srv Worker {doWork} = do
                 when (serverHostError e) $ notify c sndFileEntityId $ SFWARN e
                 liftIO $ closeXFTPServerClient c userId server digest
                 withStore' c $ \db -> updateSndChunkReplicaDelay db sndChunkReplicaId replicaDelay
-              atomically $ assertAgentForeground c
+              liftIO $ assertAgentForeground c
               loop
             retryDone e = do
               atomically $ incXFTPServerStat c userId srv uploadErrs
               sndWorkerInternalError c sndFileId sndFileEntityId (Just filePrefixPath) e
     uploadFileChunk :: AgentConfig -> SndFileChunk -> SndFileChunkReplica -> AM ()
-    uploadFileChunk AgentConfig {xftpMaxRecipientsPerRequest = maxRecipients} sndFileChunk@SndFileChunk {sndFileId, userId, chunkSpec = chunkSpec@XFTPChunkSpec {filePath}, digest = chunkDigest} replica = do
+    uploadFileChunk AgentConfig {xftpMaxRecipientsPerRequest = maxRecipients} sndFileChunk@SndFileChunk {sndFileId, userId, chunkSpec = chunkSpec@XFTPChunkSpec {filePath, chunkSize = chSize}, digest = chunkDigest} replica = do
       replica'@SndFileChunkReplica {sndChunkReplicaId} <- addRecipients sndFileChunk replica
       fsFilePath <- lift $ toFSFilePath filePath
       unlessM (doesFileExist fsFilePath) $ throwE $ FILE NO_FILE
       let chunkSpec' = chunkSpec {filePath = fsFilePath} :: XFTPChunkSpec
-      atomically $ assertAgentForeground c
+      liftIO $ assertAgentForeground c
       agentXFTPUploadChunk c userId chunkDigest replica' chunkSpec'
-      atomically $ waitUntilForeground c
+      liftIO $ waitUntilForeground c
       sf@SndFile {sndFileEntityId, prefixPath, chunks} <- withStore c $ \db -> do
         updateSndChunkReplicaStatus db sndChunkReplicaId SFRSUploaded
         getSndFile db sndFileId
@@ -521,6 +536,7 @@ runXFTPSndWorker c srv Worker {doWork} = do
           total = totalSize chunks
           complete = all chunkUploaded chunks
       atomically $ incXFTPServerStat c userId srv uploads
+      atomically $ incXFTPServerSizeStat c userId srv uploadsSize (fromIntegral $ toKB chSize)
       notify c sndFileEntityId $ SFPROG uploaded total
       when complete $ do
         (sndDescr, rcvDescrs) <- sndFileToDescrs sf
@@ -529,8 +545,8 @@ runXFTPSndWorker c srv Worker {doWork} = do
         withStore' c $ \db -> updateSndFileComplete db sndFileId
       where
         addRecipients :: SndFileChunk -> SndFileChunkReplica -> AM SndFileChunkReplica
-        addRecipients ch@SndFileChunk {numRecipients} cr@SndFileChunkReplica {rcvIdsKeys}
-          | length rcvIdsKeys > numRecipients = throwE $ INTERNAL "too many recipients"
+        addRecipients ch@SndFileChunk {numRecipients} cr@SndFileChunkReplica {sndChunkReplicaId, rcvIdsKeys}
+          | length rcvIdsKeys > numRecipients = throwE $ INTERNAL ("too many recipients, sndChunkReplicaId = " <> show sndChunkReplicaId)
           | length rcvIdsKeys == numRecipients = pure cr
           | otherwise = do
               let numRecipients' = min (numRecipients - length rcvIdsKeys) maxRecipients
@@ -650,7 +666,7 @@ runXFTPDelWorker c srv Worker {doWork} = do
   cfg <- asks config
   forever $ do
     lift $ waitForWork doWork
-    atomically $ assertAgentForeground c
+    liftIO $ assertAgentForeground c
     runXFTPOperation cfg
   where
     runXFTPOperation :: AgentConfig -> AM ()
@@ -661,6 +677,7 @@ runXFTPDelWorker c srv Worker {doWork} = do
         processDeletedReplica replica@DeletedSndChunkReplica {deletedSndChunkReplicaId, userId, server, chunkDigest, delay} = do
           let ri' = maybe ri (\d -> ri {initialInterval = d, increaseAfter = 0}) delay
           withRetryIntervalLimit xftpConsecutiveRetries ri' $ \delay' loop -> do
+            liftIO $ waitWhileSuspended c
             liftIO $ waitForUserNetwork c
             atomically $ incXFTPServerStat c userId srv deleteAttempts
             deleteChunkReplica
@@ -671,7 +688,7 @@ runXFTPDelWorker c srv Worker {doWork} = do
                 when (serverHostError e) $ notify c "" $ SFWARN e
                 liftIO $ closeXFTPServerClient c userId server chunkDigest
                 withStore' c $ \db -> updateDeletedSndChunkReplicaDelay db deletedSndChunkReplicaId replicaDelay
-              atomically $ assertAgentForeground c
+              liftIO $ assertAgentForeground c
               loop
             retryDone e = do
               atomically $ incXFTPServerStat c userId srv deleteErrs
@@ -686,7 +703,7 @@ delWorkerInternalError c deletedSndChunkReplicaId e = do
   withStore' c $ \db -> deleteDeletedSndChunkReplica db deletedSndChunkReplicaId
   notify c "" $ SFERR e
 
-assertAgentForeground :: AgentClient -> STM ()
+assertAgentForeground :: AgentClient -> IO ()
 assertAgentForeground c = do
   throwWhenInactive c
   waitUntilForeground c

src/Simplex/FileTransfer/Chunks.hs

@@ -26,6 +26,10 @@ kb :: Integral a => a -> a
 kb n = 1024 * n
 {-# INLINE kb #-}
 
+toKB :: Integral a => a -> a
+toKB n = n `div` 1024
+{-# INLINE toKB #-}
+
 mb :: Integral a => a -> a
 mb n = 1024 * kb n
 {-# INLINE mb #-}

src/Simplex/FileTransfer/Client.hs

@@ -20,15 +20,19 @@ import Data.Bifunctor (first)
 import Data.ByteString.Builder (Builder, byteString)
 import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Char8 as B
+import qualified Data.ByteString.Lazy as LB
 import Data.Int (Int64)
+import Data.List (foldl')
 import Data.List.NonEmpty (NonEmpty (..))
+import Data.Maybe (listToMaybe)
+import Data.Time.Clock (UTCTime)
 import Data.Word (Word32)
 import qualified Data.X509 as X
 import qualified Data.X509.Validation as XV
 import qualified Network.HTTP.Types as N
 import qualified Network.HTTP2.Client as H
+import Simplex.FileTransfer.Chunks
 import Simplex.FileTransfer.Protocol
-import Simplex.FileTransfer.Server.Env (supportedXFTPhandshakes)
 import Simplex.FileTransfer.Transport
 import Simplex.Messaging.Client
   ( NetworkConfig (..),
@@ -36,8 +40,8 @@ import Simplex.Messaging.Client
     TransportSession,
     chooseTransportHost,
     defaultNetworkConfig,
-    proxyUsername,
     transportClientConfig,
+    clientSocksCredentials,
     unexpectedResponse,
   )
 import qualified Simplex.Messaging.Crypto as C
@@ -50,8 +54,9 @@ import Simplex.Messaging.Protocol
     ProtocolServer (..),
     RecipientId,
     SenderId,
+    pattern NoEntity,
   )
-import Simplex.Messaging.Transport (ALPN, HandshakeError (..), THandleAuth (..), THandleParams (..), TransportError (..), TransportPeer (..), supportedParameters)
+import Simplex.Messaging.Transport (ALPN, HandshakeError (..), THandleAuth (..), THandleParams (..), TransportError (..), TransportPeer (..), defaultSupportedParams)
 import Simplex.Messaging.Transport.Client (TransportClientConfig, TransportHost, alpn)
 import Simplex.Messaging.Transport.HTTP2
 import Simplex.Messaging.Transport.HTTP2.Client
@@ -97,21 +102,21 @@ defaultXFTPClientConfig =
       clientALPN = Just supportedXFTPhandshakes
     }
 
-getXFTPClient :: TransportSession FileResponse -> XFTPClientConfig -> (XFTPClient -> IO ()) -> IO (Either XFTPClientError XFTPClient)
-getXFTPClient transportSession@(_, srv, _) config@XFTPClientConfig {clientALPN, xftpNetworkConfig, serverVRange} disconnected = runExceptT $ do
-  let username = proxyUsername transportSession
+getXFTPClient :: TransportSession FileResponse -> XFTPClientConfig -> UTCTime -> (XFTPClient -> IO ()) -> IO (Either XFTPClientError XFTPClient)
+getXFTPClient transportSession@(_, srv, _) config@XFTPClientConfig {clientALPN, xftpNetworkConfig, serverVRange} proxySessTs disconnected = runExceptT $ do
+  let socksCreds = clientSocksCredentials xftpNetworkConfig proxySessTs transportSession
       ProtocolServer _ host port keyHash = srv
   useHost <- liftEither $ chooseTransportHost xftpNetworkConfig host
-  let tcConfig = (transportClientConfig xftpNetworkConfig useHost) {alpn = clientALPN}
+  let tcConfig = (transportClientConfig xftpNetworkConfig useHost False) {alpn = clientALPN}
       http2Config = xftpHTTP2Config tcConfig config
   clientVar <- newTVarIO Nothing
   let usePort = if null port then "443" else port
       clientDisconnected = readTVarIO clientVar >>= mapM_ disconnected
-  http2Client <- liftError' xftpClientError $ getVerifiedHTTP2Client (Just username) useHost usePort (Just keyHash) Nothing http2Config clientDisconnected
+  http2Client <- liftError' xftpClientError $ getVerifiedHTTP2Client socksCreds useHost usePort (Just keyHash) Nothing http2Config clientDisconnected
   let HTTP2Client {sessionId, sessionALPN} = http2Client
       v = VersionXFTP 1
       thServerVRange = versionToRange v
-      thParams0 = THandleParams {sessionId, blockSize = xftpBlockSize, thVersion = v, thServerVRange, thAuth = Nothing, implySessId = False, batch = True}
+      thParams0 = THandleParams {sessionId, blockSize = xftpBlockSize, thVersion = v, thServerVRange, thAuth = Nothing, implySessId = False, encryptBlock = Nothing, batch = True}
   logDebug $ "Client negotiated handshake protocol: " <> tshow sessionALPN
   thParams@THandleParams {thVersion} <- case sessionALPN of
     Just "xftp/1" -> xftpClientHandshakeV1 serverVRange keyHash http2Client thParams0
@@ -171,7 +176,7 @@ xftpHTTP2Config :: TransportClientConfig -> XFTPClientConfig -> HTTP2ClientConfi
 xftpHTTP2Config transportConfig XFTPClientConfig {xftpNetworkConfig = NetworkConfig {tcpConnectTimeout}} =
   defaultHTTP2ClientConfig
     { bodyHeadSize = xftpBlockSize,
-      suportedTLSParams = supportedParameters,
+      suportedTLSParams = defaultSupportedParams,
       connTimeout = tcpConnectTimeout,
       transportConfig
     }
@@ -222,7 +227,7 @@ createXFTPChunk ::
   Maybe BasicAuth ->
   ExceptT XFTPClientError IO (SenderId, NonEmpty RecipientId)
 createXFTPChunk c spKey file rcps auth_ =
-  sendXFTPCommand c spKey "" (FNEW file rcps auth_) Nothing >>= \case
+  sendXFTPCommand c spKey NoEntity (FNEW file rcps auth_) Nothing >>= \case
     (FRSndIds sId rIds, body) -> noFile body (sId, rIds)
     (r, _) -> throwE $ unexpectedResponse r
 
@@ -278,7 +283,7 @@ pingXFTP :: XFTPClient -> ExceptT XFTPClientError IO ()
 pingXFTP c@XFTPClient {thParams} = do
   t <-
     liftEither . first PCETransportError $
-      xftpEncodeTransmission thParams ("", "", FileCmd SFRecipient PING)
+      xftpEncodeTransmission thParams ("", NoEntity, FileCmd SFRecipient PING)
   (r, _) <- sendXFTPTransmission c t Nothing
   case r of
     FRPong -> pure ()
@@ -297,3 +302,41 @@ noFile HTTP2Body {bodyPart} a = case bodyPart of
 
 -- FACK :: FileCommand Recipient
 -- PING :: FileCommand Recipient
+
+singleChunkSize :: Int64 -> Maybe Word32
+singleChunkSize size' =
+  listToMaybe $ dropWhile (< chunkSize) serverChunkSizes
+  where
+    chunkSize = fromIntegral size'
+
+prepareChunkSizes :: Int64 -> [Word32]
+prepareChunkSizes size' = prepareSizes size'
+  where
+    (smallSize, bigSize)
+      | size' > size34 chunkSize3 = (chunkSize2, chunkSize3)
+      | size' > size34 chunkSize2 = (chunkSize1, chunkSize2)
+      | otherwise = (chunkSize0, chunkSize1)
+    size34 sz = (fromIntegral sz * 3) `div` 4
+    prepareSizes 0 = []
+    prepareSizes size
+      | size >= fromIntegral bigSize = replicate (fromIntegral n1) bigSize <> prepareSizes remSz
+      | size > size34 bigSize = [bigSize]
+      | otherwise = replicate (fromIntegral n2') smallSize
+      where
+        (n1, remSz) = size `divMod` fromIntegral bigSize
+        n2' = let (n2, remSz2) = (size `divMod` fromIntegral smallSize) in if remSz2 == 0 then n2 else n2 + 1
+
+prepareChunkSpecs :: FilePath -> [Word32] -> [XFTPChunkSpec]
+prepareChunkSpecs filePath chunkSizes = reverse . snd $ foldl' addSpec (0, []) chunkSizes
+  where
+    addSpec :: (Int64, [XFTPChunkSpec]) -> Word32 -> (Int64, [XFTPChunkSpec])
+    addSpec (chunkOffset, specs) sz =
+      let spec = XFTPChunkSpec {filePath, chunkOffset, chunkSize = sz}
+       in (chunkOffset + fromIntegral sz, spec : specs)
+
+getChunkDigest :: XFTPChunkSpec -> IO ByteString
+getChunkDigest XFTPChunkSpec {filePath = chunkPath, chunkOffset, chunkSize} =
+  withFile chunkPath ReadMode $ \h -> do
+    hSeek h AbsoluteSeek $ fromIntegral chunkOffset
+    chunk <- LB.hGet h (fromIntegral chunkSize)
+    pure $! LC.sha256Hash chunk

src/Simplex/FileTransfer/Client/Agent.hs

@@ -16,6 +16,7 @@ import Data.Bifunctor (first)
 import qualified Data.ByteString.Char8 as B
 import Data.Text (Text)
 import Data.Text.Encoding (decodeUtf8)
+import Data.Time.Clock (UTCTime, getCurrentTime)
 import Simplex.FileTransfer.Client
 import Simplex.Messaging.Agent.RetryInterval
 import Simplex.Messaging.Client (NetworkConfig (..), ProtocolClientError (..), temporaryClientError)
@@ -30,6 +31,7 @@ type XFTPClientVar = TMVar (Either XFTPClientAgentError XFTPClient)
 
 data XFTPClientAgent = XFTPClientAgent
   { xftpClients :: TMap XFTPServer XFTPClientVar,
+    startedAt :: UTCTime,
     config :: XFTPClientAgentConfig
   }
 
@@ -53,22 +55,23 @@ defaultXFTPClientAgentConfig =
 data XFTPClientAgentError = XFTPClientAgentError XFTPServer XFTPClientError
   deriving (Show, Exception)
 
-newXFTPAgent :: XFTPClientAgentConfig -> STM XFTPClientAgent
+newXFTPAgent :: XFTPClientAgentConfig -> IO XFTPClientAgent
 newXFTPAgent config = do
-  xftpClients <- TM.empty
-  pure XFTPClientAgent {xftpClients, config}
+  xftpClients <- TM.emptyIO
+  startedAt <- getCurrentTime
+  pure XFTPClientAgent {xftpClients, startedAt, config}
 
 type ME a = ExceptT XFTPClientAgentError IO a
 
 getXFTPServerClient :: XFTPClientAgent -> XFTPServer -> ME XFTPClient
-getXFTPServerClient XFTPClientAgent {xftpClients, config} srv = do
+getXFTPServerClient XFTPClientAgent {xftpClients, startedAt, config} srv = do
   atomically getClientVar >>= either newXFTPClient waitForXFTPClient
   where
     connectClient :: ME XFTPClient
     connectClient =
       ExceptT $
         first (XFTPClientAgentError srv)
-          <$> getXFTPClient (1, srv, Nothing) (xftpConfig config) clientDisconnected
+          <$> getXFTPClient (1, srv, Nothing) (xftpConfig config) startedAt clientDisconnected
 
     clientDisconnected :: XFTPClient -> IO ()
     clientDisconnected _ = do

src/Simplex/FileTransfer/Client/Main.hs

@@ -19,11 +19,7 @@ module Simplex.FileTransfer.Client.Main
     singleChunkSize,
     prepareChunkSizes,
     prepareChunkSpecs,
-    maxFileSize,
-    maxFileSizeHard,
-    fileSizeLen,
     getChunkDigest,
-    SentRecipientReplica (..),
   )
 where
 
@@ -34,7 +30,6 @@ import Control.Monad.Trans.Except
 import Crypto.Random (ChaChaDRG)
 import qualified Data.Attoparsec.ByteString.Char8 as A
 import Data.Bifunctor (first)
-import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Char8 as B
 import qualified Data.ByteString.Lazy.Char8 as LB
 import Data.Char (toLower)
@@ -43,9 +38,9 @@ import Data.Int (Int64)
 import Data.List (foldl', sortOn)
 import Data.List.NonEmpty (NonEmpty (..), nonEmpty)
 import qualified Data.List.NonEmpty as L
-import Data.Map (Map)
-import qualified Data.Map as M
-import Data.Maybe (fromMaybe, listToMaybe)
+import Data.Map.Strict (Map)
+import qualified Data.Map.Strict as M
+import Data.Maybe (fromMaybe)
 import qualified Data.Text as T
 import Data.Word (Word32)
 import GHC.Records (HasField (getField))
@@ -80,20 +75,6 @@ import UnliftIO.Directory
 xftpClientVersion :: String
 xftpClientVersion = "1.0.1"
 
--- | Soft limit for XFTP clients. Should be checked and reported to user.
-maxFileSize :: Int64
-maxFileSize = gb 1
-
-maxFileSizeStr :: String
-maxFileSizeStr = B.unpack . strEncode $ FileSize maxFileSize
-
--- | Hard internal limit for XFTP agent after which it refuses to prepare chunks.
-maxFileSizeHard :: Int64
-maxFileSizeHard = gb 5
-
-fileSizeLen :: Int64
-fileSizeLen = 8
-
 newtype CLIError = CLIError String
   deriving (Eq, Show, Exception)
 
@@ -231,16 +212,6 @@ data SentFileChunkReplica = SentFileChunkReplica
   }
   deriving (Show)
 
-data SentRecipientReplica = SentRecipientReplica
-  { chunkNo :: Int,
-    server :: XFTPServer,
-    rcvNo :: Int,
-    replicaId :: ChunkReplicaId,
-    replicaKey :: C.APrivateAuthKey,
-    digest :: FileDigest,
-    chunkSize :: FileSize Word32
-  }
-
 logCfg :: LogConfig
 logCfg = LogConfig {lc_file = Nothing, lc_stderr = True}
 
@@ -313,7 +284,7 @@ cliSendFileOpts SendOptions {filePath, outputDir, numRecipients, xftpServers, re
       pure (encPath, fdRcv, fdSnd, chunkSpecs, encSize)
     uploadFile :: TVar ChaChaDRG -> [XFTPChunkSpec] -> TVar [Int64] -> Int64 -> ExceptT CLIError IO [SentFileChunk]
     uploadFile g chunks uploadedChunks encSize = do
-      a <- atomically $ newXFTPAgent defaultXFTPClientAgentConfig
+      a <- liftIO $ newXFTPAgent defaultXFTPClientAgentConfig
       gen <- newTVarIO =<< liftIO newStdGen
       let xftpSrvs = fromMaybe defaultXFTPServers (nonEmpty xftpServers)
       srvs <- liftIO $ replicateM (length chunks) $ getXFTPServer gen xftpSrvs
@@ -414,13 +385,6 @@ cliSendFileOpts SendOptions {filePath, outputDir, numRecipients, xftpServers, re
       B.writeFile fdSndPath $ strEncode fdSnd
       pure (fdRcvPaths, fdSndPath)
 
-getChunkDigest :: XFTPChunkSpec -> IO ByteString
-getChunkDigest XFTPChunkSpec {filePath = chunkPath, chunkOffset, chunkSize} =
-  withFile chunkPath ReadMode $ \h -> do
-    hSeek h AbsoluteSeek $ fromIntegral chunkOffset
-    chunk <- LB.hGet h (fromIntegral chunkSize)
-    pure $! LC.sha256Hash chunk
-
 cliReceiveFile :: ReceiveOptions -> ExceptT CLIError IO ()
 cliReceiveFile ReceiveOptions {fileDescription, filePath, retryCount, tempPath, verbose, yes} =
   getFileDescription' fileDescription >>= receive
@@ -429,7 +393,7 @@ cliReceiveFile ReceiveOptions {fileDescription, filePath, retryCount, tempPath,
     receive (ValidFileDescription FileDescription {size, digest, key, nonce, chunks}) = do
       encPath <- getEncPath tempPath "xftp"
       createDirectory encPath
-      a <- atomically $ newXFTPAgent defaultXFTPClientAgentConfig
+      a <- liftIO $ newXFTPAgent defaultXFTPClientAgentConfig
       liftIO $ printNoNewLine "Downloading file..."
       downloadedChunks <- newTVarIO []
       let srv FileChunk {replicas} = case replicas of
@@ -494,7 +458,7 @@ cliDeleteFile DeleteOptions {fileDescription, retryCount, yes} = do
   where
     deleteFile :: ValidFileDescription 'FSender -> ExceptT CLIError IO ()
     deleteFile (ValidFileDescription FileDescription {chunks}) = do
-      a <- atomically $ newXFTPAgent defaultXFTPClientAgentConfig
+      a <- liftIO $ newXFTPAgent defaultXFTPClientAgentConfig
       forM_ chunks $ deleteFileChunk a
       liftIO $ do
         printNoNewLine "File deleted!"
@@ -536,37 +500,6 @@ getFileDescription' path =
   getFileDescription path >>= \case
     AVFD fd -> either (throwE . CLIError) pure $ checkParty fd
 
-singleChunkSize :: Int64 -> Maybe Word32
-singleChunkSize size' =
-  listToMaybe $ dropWhile (< chunkSize) serverChunkSizes
-  where
-    chunkSize = fromIntegral size'
-
-prepareChunkSizes :: Int64 -> [Word32]
-prepareChunkSizes size' = prepareSizes size'
-  where
-    (smallSize, bigSize)
-      | size' > size34 chunkSize3 = (chunkSize2, chunkSize3)
-      | size' > size34 chunkSize2 = (chunkSize1, chunkSize2)
-      | otherwise = (chunkSize0, chunkSize1)
-    size34 sz = (fromIntegral sz * 3) `div` 4
-    prepareSizes 0 = []
-    prepareSizes size
-      | size >= fromIntegral bigSize = replicate (fromIntegral n1) bigSize <> prepareSizes remSz
-      | size > size34 bigSize = [bigSize]
-      | otherwise = replicate (fromIntegral n2') smallSize
-      where
-        (n1, remSz) = size `divMod` fromIntegral bigSize
-        n2' = let (n2, remSz2) = (size `divMod` fromIntegral smallSize) in if remSz2 == 0 then n2 else n2 + 1
-
-prepareChunkSpecs :: FilePath -> [Word32] -> [XFTPChunkSpec]
-prepareChunkSpecs filePath chunkSizes = reverse . snd $ foldl' addSpec (0, []) chunkSizes
-  where
-    addSpec :: (Int64, [XFTPChunkSpec]) -> Word32 -> (Int64, [XFTPChunkSpec])
-    addSpec (chunkOffset, specs) sz =
-      let spec = XFTPChunkSpec {filePath, chunkOffset, chunkSize = sz}
-       in (chunkOffset + fromIntegral sz, spec : specs)
-
 getEncPath :: MonadIO m => Maybe FilePath -> String -> m FilePath
 getEncPath path name = (`uniqueCombine` (name <> ".encrypted")) =<< maybe (liftIO getCanonicalTemporaryDirectory) pure path
 

src/Simplex/FileTransfer/Description.hs

@@ -1,7 +1,9 @@
 {-# LANGUAGE DataKinds #-}
 {-# LANGUAGE DeriveAnyClass #-}
+{-# LANGUAGE DerivingStrategies #-}
 {-# LANGUAGE DuplicateRecordFields #-}
 {-# LANGUAGE GADTs #-}
+{-# LANGUAGE GeneralizedNewtypeDeriving #-}
 {-# LANGUAGE KindSignatures #-}
 {-# LANGUAGE NamedFieldPuns #-}
 {-# LANGUAGE OverloadedStrings #-}
@@ -35,6 +37,10 @@ module Simplex.FileTransfer.Description
     FileClientData,
     fileDescriptionURI,
     qrSizeLimit,
+    maxFileSize,
+    maxFileSizeStr,
+    maxFileSizeHard,
+    fileSizeLen,
   )
 where
 
@@ -52,8 +58,8 @@ import Data.Int (Int64)
 import Data.List (foldl', sortOn)
 import Data.List.NonEmpty (NonEmpty (..))
 import qualified Data.List.NonEmpty as L
-import Data.Map (Map)
-import qualified Data.Map as M
+import Data.Map.Strict (Map)
+import qualified Data.Map.Strict as M
 import Data.Maybe (fromMaybe)
 import Data.String
 import Data.Text (Text)
@@ -139,12 +145,9 @@ data FileChunkReplica = FileChunkReplica
   }
   deriving (Eq, Show)
 
-newtype ChunkReplicaId = ChunkReplicaId {unChunkReplicaId :: ByteString}
+newtype ChunkReplicaId = ChunkReplicaId {unChunkReplicaId :: XFTPFileId}
   deriving (Eq, Show)
-
-instance StrEncoding ChunkReplicaId where
-  strEncode (ChunkReplicaId fid) = strEncode fid
-  strP = ChunkReplicaId <$> strP
+  deriving newtype (StrEncoding)
 
 instance FromJSON ChunkReplicaId where
   parseJSON = strParseJSON "ChunkReplicaId"
@@ -153,10 +156,6 @@ instance ToJSON ChunkReplicaId where
   toJSON = strToJSON
   toEncoding = strToJEncoding
 
-instance FromField ChunkReplicaId where fromField f = ChunkReplicaId <$> fromField f
-
-instance ToField ChunkReplicaId where toField (ChunkReplicaId s) = toField s
-
 data YAMLFileDescription = YAMLFileDescription
   { party :: FileParty,
     size :: String,
@@ -271,6 +270,21 @@ instance StrEncoding FileDescriptionURI where
 qrSizeLimit :: Int
 qrSizeLimit = 1002 -- ~2 chunks in URLencoded YAML with some spare size for server hosts
 
+-- | Soft limit for XFTP clients. Should be checked and reported to user.
+maxFileSize :: Int64
+maxFileSize = gb 1
+
+maxFileSizeStr :: String
+maxFileSizeStr = B.unpack . strEncode $ FileSize maxFileSize
+
+-- | Hard internal limit for XFTP agent after which it refuses to prepare chunks.
+maxFileSizeHard :: Int64
+maxFileSizeHard = gb 5
+
+fileSizeLen :: Int64
+fileSizeLen = 8
+
+
 instance (Integral a, Show a) => StrEncoding (FileSize a) where
   strEncode (FileSize b)
     | b' /= 0 = bshow b

src/Simplex/FileTransfer/Protocol.hs

@@ -41,6 +41,7 @@ import Simplex.Messaging.Protocol
     ProtocolType (..),
     RcvPublicAuthKey,
     RcvPublicDhKey,
+    EntityId (..),
     RecipientId,
     SenderId,
     SentRawTransmission,
@@ -170,7 +171,7 @@ data FileInfo = FileInfo
   }
   deriving (Show)
 
-type XFTPFileId = ByteString
+type XFTPFileId = EntityId
 
 instance FilePartyI p => ProtocolEncoding XFTPVersion XFTPErrorType (FileCommand p) where
   type Tag (FileCommand p) = FileCommandTag p
@@ -191,7 +192,7 @@ instance FilePartyI p => ProtocolEncoding XFTPVersion XFTPErrorType (FileCommand
   fromProtocolError = fromProtocolError @XFTPVersion @XFTPErrorType @FileResponse
   {-# INLINE fromProtocolError #-}
 
-  checkCredentials (auth, _, fileId, _) cmd = case cmd of
+  checkCredentials (auth, _, EntityId fileId, _) cmd = case cmd of
     -- FNEW must not have signature and chunk ID
     FNEW {}
       | isNothing auth -> Left $ CMD NO_AUTH
@@ -301,7 +302,7 @@ instance ProtocolEncoding XFTPVersion XFTPErrorType FileResponse where
     PEBlock -> BLOCK
   {-# INLINE fromProtocolError #-}
 
-  checkCredentials (_, _, entId, _) cmd = case cmd of
+  checkCredentials (_, _, EntityId entId, _) cmd = case cmd of
     FRSndIds {} -> noEntity
     -- ERR response does not always have entity ID
     FRErr _ -> Right cmd

src/Simplex/FileTransfer/Server.hs

@@ -9,6 +9,7 @@
 {-# LANGUAGE NumericUnderscores #-}
 {-# LANGUAGE OverloadedLists #-}
 {-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE PatternSynonyms #-}
 {-# LANGUAGE ScopedTypeVariables #-}
 {-# LANGUAGE TupleSections #-}
 
@@ -32,11 +33,11 @@ import qualified Data.Map.Strict as M
 import Data.Maybe (fromMaybe, isJust)
 import qualified Data.Text as T
 import Data.Time.Clock (UTCTime (..), diffTimeToPicoseconds, getCurrentTime)
-import Data.Time.Clock.System (SystemTime (..), getSystemTime)
 import Data.Time.Format.ISO8601 (iso8601Show)
 import Data.Word (Word32)
 import qualified Data.X509 as X
 import GHC.IO.Handle (hSetNewlineMode)
+import GHC.IORef (atomicSwapIORef)
 import GHC.Stats (getRTSStats)
 import qualified Network.HTTP.Types as N
 import qualified Network.HTTP2.Server as H
@@ -52,18 +53,20 @@ import qualified Simplex.Messaging.Crypto as C
 import qualified Simplex.Messaging.Crypto.Lazy as LC
 import Simplex.Messaging.Encoding
 import Simplex.Messaging.Encoding.String
-import Simplex.Messaging.Protocol (CorrId (..), RcvPublicAuthKey, RcvPublicDhKey, RecipientId, TransmissionAuth)
+import Simplex.Messaging.Protocol (CorrId (..), EntityId (..), RcvPublicAuthKey, RcvPublicDhKey, RecipientId, TransmissionAuth, pattern NoEntity)
 import Simplex.Messaging.Server (dummyVerifyCmd, verifyCmdAuthorization)
+import Simplex.Messaging.Server.Control (CPClientRole (..))
 import Simplex.Messaging.Server.Expiration
+import Simplex.Messaging.Server.QueueStore (RoundedSystemTime, getRoundedSystemTime)
 import Simplex.Messaging.Server.Stats
 import Simplex.Messaging.TMap (TMap)
 import qualified Simplex.Messaging.TMap as TM
-import Simplex.Messaging.Transport (SessionId, THandleAuth (..), THandleParams (..), TransportPeer (..))
+import Simplex.Messaging.Transport (ALPN, SessionId, THandleAuth (..), THandleParams (..), TransportPeer (..), defaultSupportedParams)
 import Simplex.Messaging.Transport.Buffer (trimCR)
 import Simplex.Messaging.Transport.HTTP2
 import Simplex.Messaging.Transport.HTTP2.File (fileBlockSize)
 import Simplex.Messaging.Transport.HTTP2.Server
-import Simplex.Messaging.Transport.Server (runTCPServer, tlsServerCredentials)
+import Simplex.Messaging.Transport.Server (runLocalTCPServer)
 import Simplex.Messaging.Util
 import Simplex.Messaging.Version
 import System.Exit (exitFailure)
@@ -89,36 +92,35 @@ data XFTPTransportRequest = XFTPTransportRequest
 runXFTPServer :: XFTPServerConfig -> IO ()
 runXFTPServer cfg = do
   started <- newEmptyTMVarIO
-  runXFTPServerBlocking started cfg
+  runXFTPServerBlocking started cfg $ Just supportedXFTPhandshakes
 
-runXFTPServerBlocking :: TMVar Bool -> XFTPServerConfig -> IO ()
-runXFTPServerBlocking started cfg = newXFTPServerEnv cfg >>= runReaderT (xftpServer cfg started)
+runXFTPServerBlocking :: TMVar Bool -> XFTPServerConfig -> Maybe [ALPN] -> IO ()
+runXFTPServerBlocking started cfg alpn_ = newXFTPServerEnv cfg >>= runReaderT (xftpServer cfg started alpn_)
 
 data Handshake
   = HandshakeSent C.PrivateKeyX25519
   | HandshakeAccepted (THandleParams XFTPVersion 'TServer)
 
-xftpServer :: XFTPServerConfig -> TMVar Bool -> M ()
-xftpServer cfg@XFTPServerConfig {xftpPort, transportConfig, inactiveClientExpiration, fileExpiration, xftpServerVRange} started = do
+xftpServer :: XFTPServerConfig -> TMVar Bool -> Maybe [ALPN] -> M ()
+xftpServer cfg@XFTPServerConfig {xftpPort, transportConfig, inactiveClientExpiration, fileExpiration, xftpServerVRange} started alpn_ = do
   mapM_ (expireServerFiles Nothing) fileExpiration
   restoreServerStats
   raceAny_ (runServer : expireFilesThread_ cfg <> serverStatsThread_ cfg <> controlPortThread_ cfg) `finally` stopServer
   where
     runServer :: M ()
     runServer = do
-      serverParams <- asks tlsServerParams
-      let (chain, pk) = tlsServerCredentials serverParams
+      srvCreds@(chain, pk) <- asks tlsServerCreds
       signKey <- liftIO $ case C.x509ToPrivate (pk, []) >>= C.privKey of
         Right pk' -> pure pk'
         Left e -> putStrLn ("servers has no valid key: " <> show e) >> exitFailure
       env <- ask
-      sessions <- atomically TM.empty
+      sessions <- liftIO TM.emptyIO
       let cleanup sessionId = atomically $ TM.delete sessionId sessions
-      liftIO . runHTTP2Server started xftpPort defaultHTTP2BufferSize serverParams transportConfig inactiveClientExpiration cleanup $ \sessionId sessionALPN r sendResponse -> do
+      liftIO . runHTTP2Server started xftpPort defaultHTTP2BufferSize defaultSupportedParams srvCreds alpn_ transportConfig inactiveClientExpiration cleanup $ \sessionId sessionALPN r sendResponse -> do
         reqBody <- getHTTP2Body r xftpBlockSize
         let v = VersionXFTP 1
             thServerVRange = versionToRange v
-            thParams0 = THandleParams {sessionId, blockSize = xftpBlockSize, thVersion = v, thServerVRange, thAuth = Nothing, implySessId = False, batch = True}
+            thParams0 = THandleParams {sessionId, blockSize = xftpBlockSize, thVersion = v, thServerVRange, thAuth = Nothing, implySessId = False, encryptBlock = Nothing, batch = True}
             req0 = XFTPTransportRequest {thParams = thParams0, request = r, reqBody, sendResponse}
         flip runReaderT env $ case sessionALPN of
           Nothing -> processRequest req0
@@ -179,6 +181,7 @@ xftpServer cfg@XFTPServerConfig {xftpPort, transportConfig, inactiveClientExpira
     stopServer = do
       withFileLog closeStoreLog
       saveServerStats
+      logInfo "Server stopped"
 
     expireFilesThread_ :: XFTPServerConfig -> [M ()]
     expireFilesThread_ XFTPServerConfig {fileExpiration = Just fileExp} = [expireFiles fileExp]
@@ -207,17 +210,17 @@ xftpServer cfg@XFTPServerConfig {xftpPort, transportConfig, inactiveClientExpira
         withFile statsFilePath AppendMode $ \h -> liftIO $ do
           hSetBuffering h LineBuffering
           ts <- getCurrentTime
-          fromTime' <- atomically $ swapTVar fromTime ts
-          filesCreated' <- atomically $ swapTVar filesCreated 0
-          fileRecipients' <- atomically $ swapTVar fileRecipients 0
-          filesUploaded' <- atomically $ swapTVar filesUploaded 0
-          filesExpired' <- atomically $ swapTVar filesExpired 0
-          filesDeleted' <- atomically $ swapTVar filesDeleted 0
-          files <- atomically $ periodStatCounts filesDownloaded ts
-          fileDownloads' <- atomically $ swapTVar fileDownloads 0
-          fileDownloadAcks' <- atomically $ swapTVar fileDownloadAcks 0
-          filesCount' <- readTVarIO filesCount
-          filesSize' <- readTVarIO filesSize
+          fromTime' <- atomicSwapIORef fromTime ts
+          filesCreated' <- atomicSwapIORef filesCreated 0
+          fileRecipients' <- atomicSwapIORef fileRecipients 0
+          filesUploaded' <- atomicSwapIORef filesUploaded 0
+          filesExpired' <- atomicSwapIORef filesExpired 0
+          filesDeleted' <- atomicSwapIORef filesDeleted 0
+          files <- liftIO $ periodStatCounts filesDownloaded ts
+          fileDownloads' <- atomicSwapIORef fileDownloads 0
+          fileDownloadAcks' <- atomicSwapIORef fileDownloadAcks 0
+          filesCount' <- readIORef filesCount
+          filesSize' <- readIORef filesSize
           hPutStrLn h $
             intercalate
               ","
@@ -247,7 +250,7 @@ xftpServer cfg@XFTPServerConfig {xftpPort, transportConfig, inactiveClientExpira
       u <- askUnliftIO
       liftIO $ do
         labelMyThread "control port server"
-        runTCPServer cpStarted port $ runCPClient u
+        runLocalTCPServer cpStarted port $ runCPClient u
       where
         runCPClient :: UnliftIO (ReaderT XFTPEnv IO) -> Socket -> IO ()
         runCPClient u sock = do
@@ -309,7 +312,7 @@ data ServerFile = ServerFile
 
 processRequest :: XFTPTransportRequest -> M ()
 processRequest XFTPTransportRequest {thParams, reqBody = body@HTTP2Body {bodyHead}, sendResponse}
-  | B.length bodyHead /= xftpBlockSize = sendXFTPResponse ("", "", FRErr BLOCK) Nothing
+  | B.length bodyHead /= xftpBlockSize = sendXFTPResponse ("", NoEntity, FRErr BLOCK) Nothing
   | otherwise = do
       case xftpDecodeTransmission thParams bodyHead of
         Right (sig_, signed, (corrId, fId, cmdOrErr)) ->
@@ -322,7 +325,7 @@ processRequest XFTPTransportRequest {thParams, reqBody = body@HTTP2Body {bodyHea
             Left e -> send (FRErr e) Nothing
           where
             send resp = sendXFTPResponse (corrId, fId, resp)
-        Left e -> sendXFTPResponse ("", "", FRErr e) Nothing
+        Left e -> sendXFTPResponse ("", NoEntity, FRErr e) Nothing
   where
     sendXFTPResponse (corrId, fId, resp) serverFile_ = do
       let t_ = xftpEncodeTransmission thParams (corrId, fId, resp)
@@ -397,7 +400,7 @@ processXFTPRequest HTTP2Body {bodyPart} = \case
       r <- runExceptT $ do
         sizes <- asks $ allowedChunkSizes . config
         unless (size file `elem` sizes) $ throwE SIZE
-        ts <- liftIO getSystemTime
+        ts <- liftIO getFileTime
         -- TODO validate body empty
         sId <- ExceptT $ addFileRetry st file 3 ts
         rcps <- mapM (ExceptT . addRecipientRetry st 3 sId) rks
@@ -405,12 +408,12 @@ processXFTPRequest HTTP2Body {bodyPart} = \case
           logAddFile sl sId file ts
           logAddRecipients sl sId rcps
         stats <- asks serverStats
-        atomically $ modifyTVar' (filesCreated stats) (+ 1)
-        atomically $ modifyTVar' (fileRecipients stats) (+ length rks)
+        lift $ incFileStat filesCreated
+        liftIO $ atomicModifyIORef'_ (fileRecipients stats) (+ length rks)
         let rIds = L.map (\(FileRecipient rId _) -> rId) rcps
         pure $ FRSndIds sId rIds
       pure $ either FRErr id r
-    addFileRetry :: FileStore -> FileInfo -> Int -> SystemTime -> M (Either XFTPErrorType XFTPFileId)
+    addFileRetry :: FileStore -> FileInfo -> Int -> RoundedSystemTime -> M (Either XFTPErrorType XFTPFileId)
     addFileRetry st file n ts =
       retryAdd n $ \sId -> runExceptT $ do
         ExceptT $ addFile st sId file ts
@@ -435,7 +438,7 @@ processXFTPRequest HTTP2Body {bodyPart} = \case
         rcps <- mapM (ExceptT . addRecipientRetry st 3 sId) rks
         lift $ withFileLog $ \sl -> logAddRecipients sl sId rcps
         stats <- asks serverStats
-        atomically $ modifyTVar' (fileRecipients stats) (+ length rks)
+        liftIO $ atomicModifyIORef'_ (fileRecipients stats) (+ length rks)
         let rIds = L.map (\(FileRecipient rId _) -> rId) rcps
         pure $ FRRcvIds rIds
       pure $ either FRErr id r
@@ -463,19 +466,19 @@ processXFTPRequest HTTP2Body {bodyPart} = \case
               \used -> let used' = used + fromIntegral size in if used' <= quota then (True, used') else (False, used)
           receive = do
             path <- asks $ filesPath . config
-            let fPath = path </> B.unpack (B64.encode senderId)
+            let fPath = path </> B.unpack (B64.encode $ unEntityId senderId)
             receiveChunk (XFTPRcvChunkSpec fPath size digest) >>= \case
               Right () -> do
                 stats <- asks serverStats
                 withFileLog $ \sl -> logPutFile sl senderId fPath
                 atomically $ writeTVar filePath (Just fPath)
-                atomically $ modifyTVar' (filesUploaded stats) (+ 1)
-                atomically $ modifyTVar' (filesCount stats) (+ 1)
-                atomically $ modifyTVar' (filesSize stats) (+ fromIntegral size)
+                incFileStat filesUploaded
+                incFileStat filesCount
+                liftIO $ atomicModifyIORef'_ (filesSize stats) (+ fromIntegral size)
                 pure FROk
               Left e -> do
                 us <- asks $ usedStorage . store
-                atomically . modifyTVar' us $ subtract (fromIntegral size)
+                atomically $ modifyTVar' us $ subtract (fromIntegral size)
                 liftIO $ whenM (doesFileExist fPath) (removeFile fPath) `catch` logFileError
                 pure $ FRErr e
           receiveChunk spec = do
@@ -494,8 +497,8 @@ processXFTPRequest HTTP2Body {bodyPart} = \case
               case LC.cbInit dhSecret cbNonce of
                 Right sbState -> do
                   stats <- asks serverStats
-                  atomically $ modifyTVar' (fileDownloads stats) (+ 1)
-                  atomically $ updatePeriodStats (filesDownloaded stats) senderId
+                  incFileStat fileDownloads
+                  liftIO $ updatePeriodStats (filesDownloaded stats) senderId
                   pure (FRFile sDhKey cbNonce, Just ServerFile {filePath = path, fileSize = size, sbState})
                 _ -> pure (FRErr INTERNAL, Nothing)
         _ -> pure (FRErr NO_FILE, Nothing)
@@ -511,8 +514,7 @@ processXFTPRequest HTTP2Body {bodyPart} = \case
       withFileLog (`logAckFile` rId)
       st <- asks store
       atomically $ deleteRecipient st rId fr
-      stats <- asks serverStats
-      atomically $ modifyTVar' (fileDownloadAcks stats) (+ 1)
+      incFileStat fileDownloadAcks
       pure FROk
 
 deleteServerFile_ :: FileRec -> M (Either XFTPErrorType ())
@@ -524,11 +526,14 @@ deleteServerFile_ FileRec {senderId, fileInfo, filePath} = do
     ExceptT $ first (\(_ :: SomeException) -> FILE_IO) <$> try (forM_ path $ \p -> whenM (doesFileExist p) (removeFile p >> deletedStats stats))
     st <- asks store
     void $ atomically $ deleteFile st senderId
-    atomically $ modifyTVar' (filesDeleted stats) (+ 1)
+    lift $ incFileStat filesDeleted
   where
     deletedStats stats = do
-      atomically $ modifyTVar' (filesCount stats) (subtract 1)
-      atomically $ modifyTVar' (filesSize stats) (subtract $ fromIntegral $ size fileInfo)
+      liftIO $ atomicModifyIORef'_ (filesCount stats) (subtract 1)
+      liftIO $ atomicModifyIORef'_ (filesSize stats) (subtract $ fromIntegral $ size fileInfo)
+
+getFileTime :: IO RoundedSystemTime
+getFileTime = getRoundedSystemTime fileTimePrecision
 
 expireServerFiles :: Maybe Int -> ExpirationConfig -> M ()
 expireServerFiles itemDelay expCfg = do
@@ -554,29 +559,26 @@ expireServerFiles itemDelay expCfg = do
     delete st sId = do
       withFileLog (`logDeleteFile` sId)
       void . atomically $ deleteFile st sId -- will not update usedStorage if sId isn't in store
-      FileServerStats {filesExpired} <- asks serverStats
-      atomically $ modifyTVar' filesExpired (+ 1)
+      incFileStat filesExpired
 
 randomId :: Int -> M ByteString
 randomId n = atomically . C.randomBytes n =<< asks random
 
 getFileId :: M XFTPFileId
-getFileId = do
-  size <- asks (fileIdSize . config)
-  atomically . C.randomBytes size =<< asks random
+getFileId = fmap EntityId . randomId =<< asks (fileIdSize . config)
 
 withFileLog :: (StoreLog 'WriteMode -> IO a) -> M ()
 withFileLog action = liftIO . mapM_ action =<< asks storeLog
 
-incFileStat :: (FileServerStats -> TVar Int) -> M ()
+incFileStat :: (FileServerStats -> IORef Int) -> M ()
 incFileStat statSel = do
   stats <- asks serverStats
-  atomically $ modifyTVar (statSel stats) (+ 1)
+  liftIO $ atomicModifyIORef'_ (statSel stats) (+ 1)
 
 saveServerStats :: M ()
 saveServerStats =
   asks (serverStatsBackupFile . config)
-    >>= mapM_ (\f -> asks serverStats >>= atomically . getFileServerStatsData >>= liftIO . saveStats f)
+    >>= mapM_ (\f -> asks serverStats >>= liftIO . getFileServerStatsData >>= liftIO . saveStats f)
   where
     saveStats f stats = do
       logInfo $ "saving server stats to file " <> T.pack f
@@ -594,7 +596,7 @@ restoreServerStats = asks (serverStatsBackupFile . config) >>= mapM_ restoreStat
           FileStore {files, usedStorage} <- asks store
           _filesCount <- M.size <$> readTVarIO files
           _filesSize <- readTVarIO usedStorage
-          atomically $ setFileServerStats s d {_filesCount, _filesSize}
+          liftIO $ setFileServerStats s d {_filesCount, _filesSize}
           renameFile f $ f <> ".bak"
           logInfo "server stats restored"
           when (statsFilesCount /= _filesCount) $ logWarn $ "Files count differs: stats: " <> tshow statsFilesCount <> ", store: " <> tshow _filesCount

src/Simplex/FileTransfer/Server/Control.hs

@@ -4,16 +4,14 @@
 module Simplex.FileTransfer.Server.Control where
 
 import qualified Data.Attoparsec.ByteString.Char8 as A
-import Data.ByteString (ByteString)
+import Simplex.FileTransfer.Protocol (XFTPFileId)
 import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Protocol (BasicAuth)
 
-data CPClientRole = CPRNone | CPRUser | CPRAdmin
-
 data ControlProtocol
   = CPAuth BasicAuth
   | CPStatsRTS
-  | CPDelete ByteString
+  | CPDelete XFTPFileId
   | CPHelp
   | CPQuit
   | CPSkip

src/Simplex/FileTransfer/Server/Env.hs

@@ -11,7 +11,6 @@ module Simplex.FileTransfer.Server.Env where
 
 import Control.Logger.Simple
 import Control.Monad
-import Control.Monad.IO.Unlift
 import Crypto.Random
 import Data.Int (Int64)
 import Data.List.NonEmpty (NonEmpty)
@@ -29,8 +28,7 @@ import Simplex.FileTransfer.Transport (VersionRangeXFTP)
 import qualified Simplex.Messaging.Crypto as C
 import Simplex.Messaging.Protocol (BasicAuth, RcvPublicAuthKey)
 import Simplex.Messaging.Server.Expiration
-import Simplex.Messaging.Transport (ALPN)
-import Simplex.Messaging.Transport.Server (TransportServerConfig (..), loadFingerprint, loadTLSServerParams)
+import Simplex.Messaging.Transport.Server (ServerCredentials (..), TransportServerConfig (..), loadFingerprint, loadServerCredential)
 import Simplex.Messaging.Util (tshow)
 import System.IO (IOMode (..))
 import UnliftIO.STM
@@ -58,10 +56,7 @@ data XFTPServerConfig = XFTPServerConfig
     fileTimeout :: Int,
     -- | time after which inactive clients can be disconnected and check interval, seconds
     inactiveClientExpiration :: Maybe ExpirationConfig,
-    -- CA certificate private key is not needed for initialization
-    caCertificateFile :: FilePath,
-    privateKeyFile :: FilePath,
-    certificateFile :: FilePath,
+    xftpCredentials :: ServerCredentials,
     -- | XFTP client-server protocol version range
     xftpServerVRange :: VersionRangeXFTP,
     -- stats config - see SMP server config
@@ -76,7 +71,7 @@ data XFTPServerConfig = XFTPServerConfig
 defaultInactiveClientExpiration :: ExpirationConfig
 defaultInactiveClientExpiration =
   ExpirationConfig
-    { ttl = 43200, -- seconds, 12 hours
+    { ttl = 21600, -- seconds, 6 hours
       checkInterval = 3600 -- seconds, 1 hours
     }
 
@@ -86,7 +81,7 @@ data XFTPEnv = XFTPEnv
     storeLog :: Maybe (StoreLog 'WriteMode),
     random :: TVar ChaChaDRG,
     serverIdentity :: C.KeyHash,
-    tlsServerParams :: T.ServerParams,
+    tlsServerCreds :: T.Credential,
     serverStats :: FileServerStats
   }
 
@@ -100,23 +95,20 @@ defaultFileExpiration =
       checkInterval = 2 * 3600 -- seconds, 2 hours
     }
 
-supportedXFTPhandshakes :: [ALPN]
-supportedXFTPhandshakes = ["xftp/1"]
-
 newXFTPServerEnv :: XFTPServerConfig -> IO XFTPEnv
-newXFTPServerEnv config@XFTPServerConfig {storeLogFile, fileSizeQuota, caCertificateFile, certificateFile, privateKeyFile, transportConfig} = do
-  random <- liftIO C.newRandom
-  store <- atomically newFileStore
-  storeLog <- liftIO $ mapM (`readWriteFileStore` store) storeLogFile
+newXFTPServerEnv config@XFTPServerConfig {storeLogFile, fileSizeQuota, xftpCredentials} = do
+  random <- C.newRandom
+  store <- newFileStore
+  storeLog <- mapM (`readWriteFileStore` store) storeLogFile
   used <- countUsedStorage <$> readTVarIO (files store)
   atomically $ writeTVar (usedStorage store) used
   forM_ fileSizeQuota $ \quota -> do
     logInfo $ "Total / available storage: " <> tshow quota <> " / " <> tshow (quota - used)
     when (quota < used) $ logInfo "WARNING: storage quota is less than used storage, no files can be uploaded!"
-  tlsServerParams <- liftIO $ loadTLSServerParams caCertificateFile certificateFile privateKeyFile (alpn transportConfig)
-  Fingerprint fp <- liftIO $ loadFingerprint caCertificateFile
-  serverStats <- atomically . newFileServerStats =<< liftIO getCurrentTime
-  pure XFTPEnv {config, store, storeLog, random, tlsServerParams, serverIdentity = C.KeyHash fp, serverStats}
+  tlsServerCreds <- loadServerCredential xftpCredentials
+  Fingerprint fp <- loadFingerprint xftpCredentials
+  serverStats <- newFileServerStats =<< getCurrentTime
+  pure XFTPEnv {config, store, storeLog, random, tlsServerCreds, serverIdentity = C.KeyHash fp, serverStats}
 
 countUsedStorage :: M.Map k FileRec -> Int64
 countUsedStorage = M.foldl' (\acc FileRec {fileInfo = FileInfo {size}} -> acc + fromIntegral size) 0

src/Simplex/FileTransfer/Server/Main.hs

@@ -19,7 +19,7 @@ import Options.Applicative
 import Simplex.FileTransfer.Chunks
 import Simplex.FileTransfer.Description (FileSize (..))
 import Simplex.FileTransfer.Server (runXFTPServer)
-import Simplex.FileTransfer.Server.Env (XFTPServerConfig (..), defFileExpirationHours, defaultFileExpiration, defaultInactiveClientExpiration, supportedXFTPhandshakes)
+import Simplex.FileTransfer.Server.Env (XFTPServerConfig (..), defFileExpirationHours, defaultFileExpiration, defaultInactiveClientExpiration)
 import Simplex.FileTransfer.Transport (supportedFileServerVRange)
 import qualified Simplex.Messaging.Crypto as C
 import Simplex.Messaging.Encoding.String
@@ -28,7 +28,7 @@ import Simplex.Messaging.Server.CLI
 import Simplex.Messaging.Server.Expiration
 import Simplex.Messaging.Transport (simplexMQVersion)
 import Simplex.Messaging.Transport.Client (TransportHost (..))
-import Simplex.Messaging.Transport.Server (TransportServerConfig (..), defaultTransportServerConfig)
+import Simplex.Messaging.Transport.Server (ServerCredentials (..), TransportServerConfig (..), defaultTransportServerConfig)
 import Simplex.Messaging.Util (safeDecodeUtf8, tshow)
 import System.Directory (createDirectoryIfMissing, doesFileExist)
 import System.FilePath (combine)
@@ -51,7 +51,9 @@ xftpServerCLI cfgPath logPath = do
         True -> readIniFile iniFile >>= either exitError runServer
         _ -> exitError $ "Error: server is not initialized (" <> iniFile <> " does not exist).\nRun `" <> executableName <> " init`."
     Delete -> do
-      confirmOrExit "WARNING: deleting the server will make all queues inaccessible, because the server identity (certificate fingerprint) will change.\nTHIS CANNOT BE UNDONE!"
+      confirmOrExit
+        "WARNING: deleting the server will make all queues inaccessible, because the server identity (certificate fingerprint) will change.\nTHIS CANNOT BE UNDONE!"
+        "Server NOT deleted"
       deleteDirIfExists cfgPath
       deleteDirIfExists logPath
       putStrLn "Deleted configuration and log files"
@@ -101,6 +103,7 @@ xftpServerCLI cfgPath logPath = do
                \\n\
                \# control_port_admin_password:\n\
                \# control_port_user_password:\n\
+               \\n\
                \[TRANSPORT]\n\
                \# host is only used to print server address on start\n"
             <> ("host: " <> T.pack host <> "\n")
@@ -173,9 +176,12 @@ xftpServerCLI cfgPath logPath = do
                     { ttl = readStrictIni "INACTIVE_CLIENTS" "ttl" ini,
                       checkInterval = readStrictIni "INACTIVE_CLIENTS" "check_interval" ini
                     },
-              caCertificateFile = c caCrtFile,
-              privateKeyFile = c serverKeyFile,
-              certificateFile = c serverCrtFile,
+              xftpCredentials =
+                ServerCredentials
+                  { caCertificateFile = Just $ c caCrtFile,
+                    privateKeyFile = c serverKeyFile,
+                    certificateFile = c serverCrtFile
+                  },
               xftpServerVRange = supportedFileServerVRange,
               logStatsInterval = logStats $> 86400, -- seconds
               logStatsStartTime = 0, -- seconds from 00:00 UTC
@@ -183,8 +189,7 @@ xftpServerCLI cfgPath logPath = do
               serverStatsBackupFile = logStats $> combine logPath "file-server-stats.log",
               transportConfig =
                 defaultTransportServerConfig
-                  { logTLSErrors = fromMaybe False $ iniOnOff "TRANSPORT" "log_tls_errors" ini,
-                    alpn = Just supportedXFTPhandshakes
+                  { logTLSErrors = fromMaybe False $ iniOnOff "TRANSPORT" "log_tls_errors" ini
                   },
               responseDelay = 0
             }

src/Simplex/FileTransfer/Server/Stats.hs

@@ -7,25 +7,24 @@ module Simplex.FileTransfer.Server.Stats where
 import Control.Applicative ((<|>))
 import qualified Data.Attoparsec.ByteString.Char8 as A
 import qualified Data.ByteString.Char8 as B
+import Data.IORef
 import Data.Int (Int64)
 import Data.Time.Clock (UTCTime)
 import Simplex.Messaging.Encoding.String
-import Simplex.Messaging.Protocol (SenderId)
 import Simplex.Messaging.Server.Stats (PeriodStats, PeriodStatsData, getPeriodStatsData, newPeriodStats, setPeriodStats)
-import UnliftIO.STM
 
 data FileServerStats = FileServerStats
-  { fromTime :: TVar UTCTime,
-    filesCreated :: TVar Int,
-    fileRecipients :: TVar Int,
-    filesUploaded :: TVar Int,
-    filesExpired :: TVar Int,
-    filesDeleted :: TVar Int,
-    filesDownloaded :: PeriodStats SenderId,
-    fileDownloads :: TVar Int,
-    fileDownloadAcks :: TVar Int,
-    filesCount :: TVar Int,
-    filesSize :: TVar Int64
+  { fromTime :: IORef UTCTime,
+    filesCreated :: IORef Int,
+    fileRecipients :: IORef Int,
+    filesUploaded :: IORef Int,
+    filesExpired :: IORef Int,
+    filesDeleted :: IORef Int,
+    filesDownloaded :: PeriodStats,
+    fileDownloads :: IORef Int,
+    fileDownloadAcks :: IORef Int,
+    filesCount :: IORef Int,
+    filesSize :: IORef Int64
   }
 
 data FileServerStatsData = FileServerStatsData
@@ -35,7 +34,7 @@ data FileServerStatsData = FileServerStatsData
     _filesUploaded :: Int,
     _filesExpired :: Int,
     _filesDeleted :: Int,
-    _filesDownloaded :: PeriodStatsData SenderId,
+    _filesDownloaded :: PeriodStatsData,
     _fileDownloads :: Int,
     _fileDownloadAcks :: Int,
     _filesCount :: Int,
@@ -43,49 +42,50 @@ data FileServerStatsData = FileServerStatsData
   }
   deriving (Show)
 
-newFileServerStats :: UTCTime -> STM FileServerStats
+newFileServerStats :: UTCTime -> IO FileServerStats
 newFileServerStats ts = do
-  fromTime <- newTVar ts
-  filesCreated <- newTVar 0
-  fileRecipients <- newTVar 0
-  filesUploaded <- newTVar 0
-  filesExpired <- newTVar 0
-  filesDeleted <- newTVar 0
+  fromTime <- newIORef ts
+  filesCreated <- newIORef 0
+  fileRecipients <- newIORef 0
+  filesUploaded <- newIORef 0
+  filesExpired <- newIORef 0
+  filesDeleted <- newIORef 0
   filesDownloaded <- newPeriodStats
-  fileDownloads <- newTVar 0
-  fileDownloadAcks <- newTVar 0
-  filesCount <- newTVar 0
-  filesSize <- newTVar 0
+  fileDownloads <- newIORef 0
+  fileDownloadAcks <- newIORef 0
+  filesCount <- newIORef 0
+  filesSize <- newIORef 0
   pure FileServerStats {fromTime, filesCreated, fileRecipients, filesUploaded, filesExpired, filesDeleted, filesDownloaded, fileDownloads, fileDownloadAcks, filesCount, filesSize}
 
-getFileServerStatsData :: FileServerStats -> STM FileServerStatsData
+getFileServerStatsData :: FileServerStats -> IO FileServerStatsData
 getFileServerStatsData s = do
-  _fromTime <- readTVar $ fromTime (s :: FileServerStats)
-  _filesCreated <- readTVar $ filesCreated s
-  _fileRecipients <- readTVar $ fileRecipients s
-  _filesUploaded <- readTVar $ filesUploaded s
-  _filesExpired <- readTVar $ filesExpired s
-  _filesDeleted <- readTVar $ filesDeleted s
+  _fromTime <- readIORef $ fromTime (s :: FileServerStats)
+  _filesCreated <- readIORef $ filesCreated s
+  _fileRecipients <- readIORef $ fileRecipients s
+  _filesUploaded <- readIORef $ filesUploaded s
+  _filesExpired <- readIORef $ filesExpired s
+  _filesDeleted <- readIORef $ filesDeleted s
   _filesDownloaded <- getPeriodStatsData $ filesDownloaded s
-  _fileDownloads <- readTVar $ fileDownloads s
-  _fileDownloadAcks <- readTVar $ fileDownloadAcks s
-  _filesCount <- readTVar $ filesCount s
-  _filesSize <- readTVar $ filesSize s
+  _fileDownloads <- readIORef $ fileDownloads s
+  _fileDownloadAcks <- readIORef $ fileDownloadAcks s
+  _filesCount <- readIORef $ filesCount s
+  _filesSize <- readIORef $ filesSize s
   pure FileServerStatsData {_fromTime, _filesCreated, _fileRecipients, _filesUploaded, _filesExpired, _filesDeleted, _filesDownloaded, _fileDownloads, _fileDownloadAcks, _filesCount, _filesSize}
 
-setFileServerStats :: FileServerStats -> FileServerStatsData -> STM ()
+-- this function is not thread safe, it is used on server start only
+setFileServerStats :: FileServerStats -> FileServerStatsData -> IO ()
 setFileServerStats s d = do
-  writeTVar (fromTime (s :: FileServerStats)) $! _fromTime (d :: FileServerStatsData)
-  writeTVar (filesCreated s) $! _filesCreated d
-  writeTVar (fileRecipients s) $! _fileRecipients d
-  writeTVar (filesUploaded s) $! _filesUploaded d
-  writeTVar (filesExpired s) $! _filesExpired d
-  writeTVar (filesDeleted s) $! _filesDeleted d
+  writeIORef (fromTime (s :: FileServerStats)) $! _fromTime (d :: FileServerStatsData)
+  writeIORef (filesCreated s) $! _filesCreated d
+  writeIORef (fileRecipients s) $! _fileRecipients d
+  writeIORef (filesUploaded s) $! _filesUploaded d
+  writeIORef (filesExpired s) $! _filesExpired d
+  writeIORef (filesDeleted s) $! _filesDeleted d
   setPeriodStats (filesDownloaded s) $! _filesDownloaded d
-  writeTVar (fileDownloads s) $! _fileDownloads d
-  writeTVar (fileDownloadAcks s) $! _fileDownloadAcks d
-  writeTVar (filesCount s) $! _filesCount d
-  writeTVar (filesSize s) $! _filesSize d
+  writeIORef (fileDownloads s) $! _fileDownloads d
+  writeIORef (fileDownloadAcks s) $! _fileDownloadAcks d
+  writeIORef (filesCount s) $! _filesCount d
+  writeIORef (filesSize s) $! _filesSize d
 
 instance StrEncoding FileServerStatsData where
   strEncode FileServerStatsData {_fromTime, _filesCreated, _fileRecipients, _filesUploaded, _filesExpired, _filesDeleted, _filesDownloaded, _fileDownloads, _fileDownloadAcks, _filesCount, _filesSize} =

src/Simplex/FileTransfer/Server/Store.hs

@@ -17,6 +17,7 @@ module Simplex.FileTransfer.Server.Store
     expiredFilePath,
     getFile,
     ackFile,
+    fileTimePrecision,
   )
 where
 
@@ -25,12 +26,12 @@ import qualified Data.Attoparsec.ByteString.Char8 as A
 import Data.Int (Int64)
 import Data.Set (Set)
 import qualified Data.Set as S
-import Data.Time.Clock.System (SystemTime (..))
 import Simplex.FileTransfer.Protocol (FileInfo (..), SFileParty (..), XFTPFileId)
 import Simplex.FileTransfer.Transport (XFTPErrorType (..))
 import qualified Simplex.Messaging.Crypto as C
 import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Protocol (RcvPublicAuthKey, RecipientId, SenderId)
+import Simplex.Messaging.Server.QueueStore (RoundedSystemTime (..))
 import Simplex.Messaging.TMap (TMap)
 import qualified Simplex.Messaging.TMap as TM
 import Simplex.Messaging.Util (ifM, ($>>=))
@@ -46,30 +47,34 @@ data FileRec = FileRec
     fileInfo :: FileInfo,
     filePath :: TVar (Maybe FilePath),
     recipientIds :: TVar (Set RecipientId),
-    createdAt :: SystemTime
+    createdAt :: RoundedSystemTime
   }
 
+fileTimePrecision :: Int64
+fileTimePrecision = 3600 -- truncate creation time to 1 hour
+
 data FileRecipient = FileRecipient RecipientId RcvPublicAuthKey
+  deriving (Show)
 
 instance StrEncoding FileRecipient where
   strEncode (FileRecipient rId rKey) = strEncode rId <> ":" <> strEncode rKey
   strP = FileRecipient <$> strP <* A.char ':' <*> strP
 
-newFileStore :: STM FileStore
+newFileStore :: IO FileStore
 newFileStore = do
-  files <- TM.empty
-  recipients <- TM.empty
-  usedStorage <- newTVar 0
+  files <- TM.emptyIO
+  recipients <- TM.emptyIO
+  usedStorage <- newTVarIO 0
   pure FileStore {files, recipients, usedStorage}
 
-addFile :: FileStore -> SenderId -> FileInfo -> SystemTime -> STM (Either XFTPErrorType ())
+addFile :: FileStore -> SenderId -> FileInfo -> RoundedSystemTime -> STM (Either XFTPErrorType ())
 addFile FileStore {files} sId fileInfo createdAt =
   ifM (TM.member sId files) (pure $ Left DUPLICATE_) $ do
     f <- newFileRec sId fileInfo createdAt
     TM.insert sId f files
     pure $ Right ()
 
-newFileRec :: SenderId -> FileInfo -> SystemTime -> STM FileRec
+newFileRec :: SenderId -> FileInfo -> RoundedSystemTime -> STM FileRec
 newFileRec senderId fileInfo createdAt = do
   recipientIds <- newTVar S.empty
   filePath <- newTVar Nothing
@@ -120,8 +125,8 @@ getFile st party fId = case party of
 expiredFilePath :: FileStore -> XFTPFileId -> Int64 -> STM (Maybe (Maybe FilePath))
 expiredFilePath FileStore {files} sId old =
   TM.lookup sId files
-    $>>= \FileRec {filePath, createdAt} ->
-      if systemSeconds createdAt < old
+    $>>= \FileRec {filePath, createdAt = RoundedSystemTime createdAt} ->
+      if createdAt + fileTimePrecision < old
         then Just <$> readTVar filePath
         else pure Nothing
 

src/Simplex/FileTransfer/Server/StoreLog.hs

@@ -28,22 +28,22 @@ import Data.List.NonEmpty (NonEmpty)
 import qualified Data.List.NonEmpty as L
 import Data.Map.Strict (Map)
 import qualified Data.Map.Strict as M
-import Data.Time.Clock.System (SystemTime)
 import Simplex.FileTransfer.Protocol (FileInfo (..))
 import Simplex.FileTransfer.Server.Store
 import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Protocol (RcvPublicAuthKey, RecipientId, SenderId)
+import Simplex.Messaging.Server.QueueStore (RoundedSystemTime)
 import Simplex.Messaging.Server.StoreLog
-import Simplex.Messaging.Util (bshow, whenM)
-import System.Directory (doesFileExist, renameFile)
+import Simplex.Messaging.Util (bshow)
 import System.IO
 
 data FileStoreLogRecord
-  = AddFile SenderId FileInfo SystemTime
+  = AddFile SenderId FileInfo RoundedSystemTime
   | PutFile SenderId FilePath
   | AddRecipients SenderId (NonEmpty FileRecipient)
   | DeleteFile SenderId
   | AckFile RecipientId
+  deriving (Show)
 
 instance StrEncoding FileStoreLogRecord where
   strEncode = \case
@@ -64,7 +64,7 @@ instance StrEncoding FileStoreLogRecord where
 logFileStoreRecord :: StoreLog 'WriteMode -> FileStoreLogRecord -> IO ()
 logFileStoreRecord = writeStoreLogRecord
 
-logAddFile :: StoreLog 'WriteMode -> SenderId -> FileInfo -> SystemTime -> IO ()
+logAddFile :: StoreLog 'WriteMode -> SenderId -> FileInfo -> RoundedSystemTime -> IO ()
 logAddFile s = logFileStoreRecord s .:. AddFile
 
 logPutFile :: StoreLog 'WriteMode -> SenderId -> FilePath -> IO ()
@@ -80,13 +80,7 @@ logAckFile :: StoreLog 'WriteMode -> RecipientId -> IO ()
 logAckFile s = logFileStoreRecord s . AckFile
 
 readWriteFileStore :: FilePath -> FileStore -> IO (StoreLog 'WriteMode)
-readWriteFileStore f st = do
-  whenM (doesFileExist f) $ do
-    readFileStore f st
-    renameFile f $ f <> ".bak"
-  s <- openWriteStoreLog f
-  writeFileStore s st
-  pure s
+readWriteFileStore = readWriteStoreLog readFileStore writeFileStore
 
 readFileStore :: FilePath -> FileStore -> IO ()
 readFileStore f st = mapM_ (addFileLogRecord . LB.toStrict) . LB.lines =<< LB.readFile f

src/Simplex/FileTransfer/Transport.hs

@@ -12,6 +12,7 @@ module Simplex.FileTransfer.Transport
   ( supportedFileServerVRange,
     authCmdsXFTPVersion,
     xftpClientHandshakeStub,
+    supportedXFTPhandshakes,
     XFTPClientHandshake (..),
     -- xftpClientHandshake,
     XFTPServerHandshake (..),
@@ -57,7 +58,7 @@ import Simplex.Messaging.Encoding
 import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Parsers
 import Simplex.Messaging.Protocol (CommandError)
-import Simplex.Messaging.Transport (SessionId, THandle (..), THandleParams (..), TransportError (..), TransportPeer (..))
+import Simplex.Messaging.Transport (ALPN, SessionId, THandle (..), THandleParams (..), TransportError (..), TransportPeer (..))
 import Simplex.Messaging.Transport.HTTP2.File
 import Simplex.Messaging.Util (bshow, tshow)
 import Simplex.Messaging.Version
@@ -101,6 +102,9 @@ supportedFileServerVRange = mkVersionRange initialXFTPVersion currentXFTPVersion
 xftpClientHandshakeStub :: c -> Maybe C.KeyPairX25519 -> C.KeyHash -> VersionRangeXFTP -> ExceptT TransportError IO (THandle XFTPVersion c 'TClient)
 xftpClientHandshakeStub _c _ks _keyHash _xftpVRange = throwE TEVersion
 
+supportedXFTPhandshakes :: [ALPN]
+supportedXFTPhandshakes = ["xftp/1"]
+
 data XFTPServerHandshake = XFTPServerHandshake
   { xftpVersionRange :: VersionRangeXFTP,
     sessionId :: SessionId,

src/Simplex/FileTransfer/Types.hs

@@ -25,9 +25,9 @@ import Simplex.Messaging.Parsers
 import Simplex.Messaging.Protocol (XFTPServer)
 import System.FilePath ((</>))
 
-type RcvFileId = ByteString
+type RcvFileId = ByteString -- Agent entity ID
 
-type SndFileId = ByteString
+type SndFileId = ByteString -- Agent entity ID
 
 authTagSize :: Int64
 authTagSize = fromIntegral C.authTagSize
@@ -246,6 +246,16 @@ data DeletedSndChunkReplica = DeletedSndChunkReplica
   }
   deriving (Show)
 
+data SentRecipientReplica = SentRecipientReplica
+  { chunkNo :: Int,
+    server :: XFTPServer,
+    rcvNo :: Int,
+    replicaId :: ChunkReplicaId,
+    replicaKey :: C.APrivateAuthKey,
+    digest :: FileDigest,
+    chunkSize :: FileSize Word32
+  }
+
 data FileErrorType
   = -- | cannot proceed with download from not approved relays without proxy
     NOT_APPROVED

src/Simplex/Messaging/Agent/Env/SQLite.hs

@@ -7,6 +7,7 @@
 {-# LANGUAGE NumericUnderscores #-}
 {-# LANGUAGE RankNTypes #-}
 {-# LANGUAGE ScopedTypeVariables #-}
+{-# LANGUAGE TemplateHaskell #-}
 {-# LANGUAGE TypeApplications #-}
 {-# OPTIONS_GHC -fno-warn-unticked-promoted-constructors #-}
 
@@ -15,7 +16,15 @@ module Simplex.Messaging.Agent.Env.SQLite
     AM,
     AgentConfig (..),
     InitialAgentServers (..),
+    ServerCfg (..),
+    ServerRoles (..),
+    OperatorId,
+    UserServers (..),
     NetworkConfig (..),
+    presetServerCfg,
+    allRoles,
+    mkUserServers,
+    serverHosts,
     defaultAgentConfig,
     defaultReconnectInterval,
     tryAgentError,
@@ -35,14 +44,22 @@ module Simplex.Messaging.Agent.Env.SQLite
   )
 where
 
+import Control.Concurrent (ThreadId)
+import Control.Exception (BlockedIndefinitelyOnSTM (..), SomeException, fromException)
 import Control.Monad.Except
 import Control.Monad.IO.Unlift
 import Control.Monad.Reader
 import Crypto.Random
+import Data.Aeson (FromJSON (..), ToJSON (..))
+import qualified Data.Aeson.TH as JQ
 import Data.ByteArray (ScrubbedBytes)
 import Data.Int (Int64)
 import Data.List.NonEmpty (NonEmpty)
-import Data.Map (Map)
+import qualified Data.List.NonEmpty as L
+import Data.Map.Strict (Map)
+import Data.Maybe (fromMaybe)
+import Data.Set (Set)
+import qualified Data.Set as S
 import Data.Time.Clock (NominalDiffTime, nominalDay)
 import Data.Time.Clock.System (SystemTime (..))
 import Data.Word (Word16)
@@ -59,14 +76,15 @@ import Simplex.Messaging.Crypto.Ratchet (VersionRangeE2E, supportedE2EEncryptVRa
 import Simplex.Messaging.Notifications.Client (defaultNTFClientConfig)
 import Simplex.Messaging.Notifications.Transport (NTFVersion)
 import Simplex.Messaging.Notifications.Types
-import Simplex.Messaging.Protocol (NtfServer, VersionRangeSMPC, XFTPServer, XFTPServerWithAuth, supportedSMPClientVRange)
+import Simplex.Messaging.Parsers (defaultJSON)
+import Simplex.Messaging.Protocol (NtfServer, ProtoServerWithAuth (..), ProtocolServer (..), ProtocolType (..), ProtocolTypeI, VersionRangeSMPC, XFTPServer, supportedSMPClientVRange)
 import Simplex.Messaging.TMap (TMap)
 import qualified Simplex.Messaging.TMap as TM
-import Simplex.Messaging.Transport (SMPVersion, TLS, Transport (..))
-import Simplex.Messaging.Transport.Client (defaultSMPPort)
+import Simplex.Messaging.Transport (SMPVersion)
+import Simplex.Messaging.Transport.Client (TransportHost)
 import Simplex.Messaging.Util (allFinally, catchAllErrors, catchAllErrors', tryAllErrors, tryAllErrors')
+import System.Mem.Weak (Weak)
 import System.Random (StdGen, newStdGen)
-import UnliftIO (Async, SomeException)
 import UnliftIO.STM
 
 type AM' a = ReaderT Env IO a
@@ -74,12 +92,51 @@ type AM' a = ReaderT Env IO a
 type AM a = ExceptT AgentErrorType (ReaderT Env IO) a
 
 data InitialAgentServers = InitialAgentServers
-  { smp :: Map UserId (NonEmpty SMPServerWithAuth),
+  { smp :: Map UserId (NonEmpty (ServerCfg 'PSMP)),
     ntf :: [NtfServer],
-    xftp :: Map UserId (NonEmpty XFTPServerWithAuth),
+    xftp :: Map UserId (NonEmpty (ServerCfg 'PXFTP)),
     netCfg :: NetworkConfig
   }
 
+data ServerCfg p = ServerCfg
+  { server :: ProtoServerWithAuth p,
+    operator :: Maybe OperatorId,
+    enabled :: Bool,
+    roles :: ServerRoles
+  }
+  deriving (Show)
+
+data ServerRoles = ServerRoles
+  { storage :: Bool,
+    proxy :: Bool
+  }
+  deriving (Show)
+
+allRoles :: ServerRoles
+allRoles = ServerRoles True True
+
+presetServerCfg :: Bool -> ServerRoles -> Maybe OperatorId -> ProtoServerWithAuth p -> ServerCfg p
+presetServerCfg enabled roles operator server =
+  ServerCfg {server, operator, enabled, roles}
+
+data UserServers p = UserServers
+  { storageSrvs :: NonEmpty (Maybe OperatorId, ProtoServerWithAuth p),
+    proxySrvs :: NonEmpty (Maybe OperatorId, ProtoServerWithAuth p),
+    knownHosts :: Set TransportHost
+  }
+
+type OperatorId = Int64
+
+-- This function sets all servers as enabled in case all passed servers are disabled.
+mkUserServers :: NonEmpty (ServerCfg p) -> UserServers p
+mkUserServers srvs = UserServers {storageSrvs = filterSrvs storage, proxySrvs = filterSrvs proxy, knownHosts}
+  where
+    filterSrvs role = L.map (\ServerCfg {operator, server} -> (operator, server)) $ fromMaybe srvs $ L.nonEmpty $ L.filter (\ServerCfg {enabled, roles} -> enabled && role roles) srvs
+    knownHosts = S.unions $ L.map (\ServerCfg {server = ProtoServerWithAuth srv _} -> serverHosts srv) srvs
+
+serverHosts :: ProtocolServer p -> Set TransportHost
+serverHosts ProtocolServer {host} = S.fromList $ L.toList host
+
 data AgentConfig = AgentConfig
   { tcpPort :: Maybe ServiceName,
     rcvAuthAlg :: C.AuthAlg,
@@ -111,10 +168,9 @@ data AgentConfig = AgentConfig
     xftpMaxRecipientsPerRequest :: Int,
     deleteErrorCount :: Int,
     ntfCron :: Word16,
-    ntfWorkerDelay :: Int,
-    ntfSMPWorkerDelay :: Int,
+    ntfBatchSize :: Int,
+    ntfSubFirstCheckInterval :: NominalDiffTime,
     ntfSubCheckInterval :: NominalDiffTime,
-    ntfMaxMessages :: Int,
     caCertificateFile :: FilePath,
     privateKeyFile :: FilePath,
     certificateFile :: FilePath,
@@ -128,7 +184,7 @@ defaultReconnectInterval =
   RetryInterval
     { initialInterval = 2_000000,
       increaseAfter = 10_000000,
-      maxInterval = 60_000000
+      maxInterval = 180_000000
     }
 
 defaultMessageRetryInterval :: RetryInterval2
@@ -138,7 +194,7 @@ defaultMessageRetryInterval =
         RetryInterval
           { initialInterval = 2_000000,
             increaseAfter = 10_000000,
-            maxInterval = 60_000000
+            maxInterval = 120_000000
           },
       riSlow =
         RetryInterval
@@ -157,9 +213,9 @@ defaultAgentConfig =
       rcvAuthAlg = C.AuthAlg C.SEd25519, -- this will stay as Ed25519
       sndAuthAlg = C.AuthAlg C.SEd25519, -- TODO replace with X25519 when switching to v7
       connIdBytes = 12,
-      tbqSize = 64,
-      smpCfg = defaultSMPClientConfig {defaultTransport = (show defaultSMPPort, transport @TLS)},
-      ntfCfg = defaultNTFClientConfig {defaultTransport = ("443", transport @TLS)},
+      tbqSize = 128,
+      smpCfg = defaultSMPClientConfig,
+      ntfCfg = defaultNTFClientConfig,
       xftpCfg = defaultXFTPClientConfig,
       reconnectInterval = defaultReconnectInterval,
       messageRetryInterval = defaultMessageRetryInterval,
@@ -183,10 +239,9 @@ defaultAgentConfig =
       xftpMaxRecipientsPerRequest = 200,
       deleteErrorCount = 10,
       ntfCron = 20, -- minutes
-      ntfWorkerDelay = 100000, -- microseconds
-      ntfSMPWorkerDelay = 500000, -- microseconds
-      ntfSubCheckInterval = nominalDay,
-      ntfMaxMessages = 3,
+      ntfBatchSize = 150,
+      ntfSubFirstCheckInterval = nominalDay,
+      ntfSubCheckInterval = 3 * nominalDay,
       -- CA certificate private key is not needed for initialization
       -- ! we do not generate these
       caCertificateFile = "/etc/opt/simplex-agent/ca.crt",
@@ -211,8 +266,8 @@ newSMPAgentEnv :: AgentConfig -> SQLiteStore -> IO Env
 newSMPAgentEnv config store = do
   random <- C.newRandom
   randomServer <- newTVarIO =<< liftIO newStdGen
-  ntfSupervisor <- atomically . newNtfSubSupervisor $ tbqSize config
-  xftpAgent <- atomically newXFTPAgent
+  ntfSupervisor <- newNtfSubSupervisor $ tbqSize config
+  xftpAgent <- newXFTPAgent
   multicastSubscribers <- newTMVarIO 0
   pure Env {config, store, random, randomServer, ntfSupervisor, xftpAgent, multicastSubscribers}
 
@@ -221,21 +276,23 @@ createAgentStore dbFilePath dbKey keepKey = createSQLiteStore dbFilePath dbKey k
 
 data NtfSupervisor = NtfSupervisor
   { ntfTkn :: TVar (Maybe NtfToken),
-    ntfSubQ :: TBQueue (ConnId, NtfSupervisorCommand),
+    ntfSubQ :: TBQueue (NtfSupervisorCommand, NonEmpty ConnId),
     ntfWorkers :: TMap NtfServer Worker,
-    ntfSMPWorkers :: TMap SMPServer Worker
+    ntfSMPWorkers :: TMap SMPServer Worker,
+    ntfTknDelWorkers :: TMap NtfServer Worker
   }
 
-data NtfSupervisorCommand = NSCCreate | NSCDelete | NSCSmpDelete | NSCNtfWorker NtfServer | NSCNtfSMPWorker SMPServer
+data NtfSupervisorCommand = NSCCreate | NSCSmpDelete | NSCDeleteSub
   deriving (Show)
 
-newNtfSubSupervisor :: Natural -> STM NtfSupervisor
+newNtfSubSupervisor :: Natural -> IO NtfSupervisor
 newNtfSubSupervisor qSize = do
-  ntfTkn <- newTVar Nothing
-  ntfSubQ <- newTBQueue qSize
-  ntfWorkers <- TM.empty
-  ntfSMPWorkers <- TM.empty
-  pure NtfSupervisor {ntfTkn, ntfSubQ, ntfWorkers, ntfSMPWorkers}
+  ntfTkn <- newTVarIO Nothing
+  ntfSubQ <- newTBQueueIO qSize
+  ntfWorkers <- TM.emptyIO
+  ntfSMPWorkers <- TM.emptyIO
+  ntfTknDelWorkers <- TM.emptyIO
+  pure NtfSupervisor {ntfTkn, ntfSubQ, ntfWorkers, ntfSMPWorkers, ntfTknDelWorkers}
 
 data XFTPAgent = XFTPAgent
   { -- if set, XFTP file paths will be considered as relative to this directory
@@ -245,12 +302,12 @@ data XFTPAgent = XFTPAgent
     xftpDelWorkers :: TMap XFTPServer Worker
   }
 
-newXFTPAgent :: STM XFTPAgent
+newXFTPAgent :: IO XFTPAgent
 newXFTPAgent = do
-  xftpWorkDir <- newTVar Nothing
-  xftpRcvWorkers <- TM.empty
-  xftpSndWorkers <- TM.empty
-  xftpDelWorkers <- TM.empty
+  xftpWorkDir <- newTVarIO Nothing
+  xftpRcvWorkers <- TM.emptyIO
+  xftpSndWorkers <- TM.emptyIO
+  xftpDelWorkers <- TM.emptyIO
   pure XFTPAgent {xftpWorkDir, xftpRcvWorkers, xftpSndWorkers, xftpDelWorkers}
 
 tryAgentError :: AM a -> AM (Either AgentErrorType a)
@@ -275,13 +332,15 @@ agentFinally = allFinally mkInternal
 {-# INLINE agentFinally #-}
 
 mkInternal :: SomeException -> AgentErrorType
-mkInternal = INTERNAL . show
+mkInternal e = case fromException e of
+  Just BlockedIndefinitelyOnSTM -> CRITICAL True "Thread blocked indefinitely in STM transaction"
+  _ -> INTERNAL $ show e
 {-# INLINE mkInternal #-}
 
 data Worker = Worker
   { workerId :: Int,
     doWork :: TMVar (),
-    action :: TMVar (Maybe (Async ())),
+    action :: TMVar (Maybe (Weak ThreadId)),
     restarts :: TVar RestartCount
   }
 
@@ -294,3 +353,14 @@ updateRestartCount :: SystemTime -> RestartCount -> RestartCount
 updateRestartCount t (RestartCount minute count) = do
   let min' = systemSeconds t `div` 60
    in RestartCount min' $ if minute == min' then count + 1 else 1
+
+$(pure [])
+
+$(JQ.deriveJSON defaultJSON ''ServerRoles)
+
+instance ProtocolTypeI p => ToJSON (ServerCfg p) where
+  toEncoding = $(JQ.mkToEncoding defaultJSON ''ServerCfg)
+  toJSON = $(JQ.mkToJSON defaultJSON ''ServerCfg)
+
+instance ProtocolTypeI p => FromJSON (ServerCfg p) where
+  parseJSON = $(JQ.mkParseJSON defaultJSON ''ServerCfg)

src/Simplex/Messaging/Agent/Lock.hs

@@ -1,6 +1,7 @@
 module Simplex.Messaging.Agent.Lock
   ( Lock,
     createLock,
+    createLockIO,
     withLock,
     withLock',
     withGetLock,
@@ -24,6 +25,10 @@ createLock :: STM Lock
 createLock = newEmptyTMVar
 {-# INLINE createLock #-}
 
+createLockIO :: IO Lock
+createLockIO = newEmptyTMVarIO
+{-# INLINE createLockIO #-}
+
 withLock :: MonadUnliftIO m => Lock -> String -> ExceptT e m a -> ExceptT e m a
 withLock lock name = ExceptT . withLock' lock name . runExceptT
 {-# INLINE withLock #-}

src/Simplex/Messaging/Agent/NtfSubSupervisor.hs

@@ -1,7 +1,9 @@
+{-# LANGUAGE DataKinds #-}
 {-# LANGUAGE DuplicateRecordFields #-}
 {-# LANGUAGE FlexibleContexts #-}
 {-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE OverloadedLists #-}
 {-# LANGUAGE OverloadedStrings #-}
 {-# LANGUAGE ScopedTypeVariables #-}
 {-# LANGUAGE TupleSections #-}
@@ -13,6 +15,7 @@ module Simplex.Messaging.Agent.NtfSubSupervisor
     nsRemoveNtfToken,
     sendNtfSubCommand,
     instantNotifications,
+    deleteToken,
     closeNtfSupervisor,
     getNtfServer,
   )
@@ -20,117 +23,170 @@ where
 
 import Control.Logger.Simple (logError, logInfo)
 import Control.Monad
-import Control.Monad.Except
 import Control.Monad.Reader
+import Control.Monad.Trans.Except
+import Crypto.Random (ChaChaDRG)
 import Data.Bifunctor (first)
+import Data.Either (fromRight, partitionEithers)
+import Data.Functor (($>))
+import Data.List (foldl')
+import Data.List.NonEmpty (NonEmpty (..))
+import qualified Data.List.NonEmpty as L
 import qualified Data.Map.Strict as M
-import Data.Text (Text)
+import Data.Maybe (catMaybes)
+import qualified Data.Set as S
 import Data.Time (UTCTime, addUTCTime, getCurrentTime)
 import Data.Time.Clock (diffUTCTime)
 import Simplex.Messaging.Agent.Client
 import Simplex.Messaging.Agent.Env.SQLite
-import Simplex.Messaging.Agent.Protocol (AEvent (..), AEvt (..), AgentErrorType (..), BrokerErrorType (..), ConnId, NotificationsMode (..), SAEntity (..))
+import Simplex.Messaging.Agent.Protocol
 import Simplex.Messaging.Agent.RetryInterval
+import Simplex.Messaging.Agent.Stats
 import Simplex.Messaging.Agent.Store
 import Simplex.Messaging.Agent.Store.SQLite
+import qualified Simplex.Messaging.Agent.Store.SQLite.DB as DB
 import qualified Simplex.Messaging.Crypto as C
-import Simplex.Messaging.Notifications.Protocol (NtfSubStatus (..), NtfTknStatus (..), SMPQueueNtf (..))
+import Simplex.Messaging.Notifications.Protocol
 import Simplex.Messaging.Notifications.Types
-import Simplex.Messaging.Protocol (NtfServer, SMPServer, sameSrvAddr)
-import Simplex.Messaging.Util (diffToMicroseconds, threadDelay', tshow, unlessM)
+import Simplex.Messaging.Protocol (NtfServer, sameSrvAddr)
+import qualified Simplex.Messaging.Protocol as SMP
+import Simplex.Messaging.Util (diffToMicroseconds, threadDelay', tshow)
 import System.Random (randomR)
 import UnliftIO
-import UnliftIO.Concurrent (forkIO, threadDelay)
+import UnliftIO.Concurrent (forkIO)
 import qualified UnliftIO.Exception as E
 
 runNtfSupervisor :: AgentClient -> AM' ()
 runNtfSupervisor c = do
   ns <- asks ntfSupervisor
+  runExceptT startTknDelete >>= \case
+    Left e -> notifyErr e
+    Right _ -> pure ()
   forever $ do
-    cmd@(connId, _) <- atomically . readTBQueue $ ntfSubQ ns
-    handleErr connId . agentOperationBracket c AONtfNetwork waitUntilActive $
-      runExceptT (processNtfSub c cmd) >>= \case
-        Left e -> notifyErr connId e
+    cmd <- atomically . readTBQueue $ ntfSubQ ns
+    handleErr . agentOperationBracket c AONtfNetwork waitUntilActive $
+      runExceptT (processNtfCmd c cmd) >>= \case
+        Left e -> notifyErr e
         Right _ -> return ()
   where
-    handleErr :: ConnId -> AM' () -> AM' ()
-    handleErr connId = E.handle $ \(e :: E.SomeException) -> do
+    startTknDelete :: AM ()
+    startTknDelete = do
+      pendingDelServers <- withStore' c getPendingDelTknServers
+      lift . forM_ pendingDelServers $ getNtfTknDelWorker True c
+    handleErr :: AM' () -> AM' ()
+    handleErr = E.handle $ \(e :: E.SomeException) -> do
       logError $ "runNtfSupervisor error " <> tshow e
-      notifyErr connId e
-    notifyErr connId e = notifyInternalError c connId $ "runNtfSupervisor error " <> show e
+      notifyErr e
+    notifyErr e = notifyInternalError' c $ "runNtfSupervisor error " <> show e
 
-processNtfSub :: AgentClient -> (ConnId, NtfSupervisorCommand) -> AM ()
-processNtfSub c (connId, cmd) = do
-  logInfo $ "processNtfSub - connId = " <> tshow connId <> " - cmd = " <> tshow cmd
+partitionErrs :: (a -> ConnId) -> [a] -> [Either AgentErrorType b] -> ([(ConnId, AgentErrorType)], [b])
+partitionErrs f xs = partitionEithers . zipWith (\x -> first (f x,)) xs
+{-# INLINE partitionErrs #-}
+
+ntfSubConnId :: NtfSubscription -> ConnId
+ntfSubConnId NtfSubscription {connId} = connId
+
+processNtfCmd :: AgentClient -> (NtfSupervisorCommand, NonEmpty ConnId) -> AM ()
+processNtfCmd c (cmd, connIds) = do
+  logInfo $ "processNtfCmd - cmd = " <> tshow cmd
+  let connIds' = L.toList connIds
   case cmd of
     NSCCreate -> do
-      (a, RcvQueue {server = smpServer, clientNtfCreds}) <- withStore c $ \db -> runExceptT $ do
-        a <- liftIO $ getNtfSubscription db connId
-        q <- ExceptT $ getPrimaryRcvQueue db connId
-        pure (a, q)
-      logInfo $ "processNtfSub, NSCCreate - a = " <> tshow a
-      case a of
-        Nothing -> do
+      (cErrs, rqSubActions) <- lift $ partitionErrs id connIds' <$> withStoreBatch c (\db -> map (getQueueSub db) connIds')
+      notifyErrs c cErrs
+      logInfo $ "processNtfCmd, NSCCreate - length rqSubs = " <> tshow (length rqSubActions)
+      let (ns, rs, css, cns) = partitionQueueSubActions rqSubActions
+      createNewSubs ns
+      resetSubs rs
+      lift $ do
+        mapM_ (getNtfSMPWorker True c) (S.fromList css)
+        mapM_ (getNtfNTFWorker True c) (S.fromList cns)
+      where
+        getQueueSub ::
+          DB.Connection ->
+          ConnId ->
+          IO (Either AgentErrorType (RcvQueue, Maybe NtfSupervisorSub))
+        getQueueSub db connId = fmap (first storeError) $ runExceptT $ do
+          rq <- ExceptT $ getPrimaryRcvQueue db connId
+          sub <- liftIO $ getNtfSubscription db connId
+          pure (rq, sub)
+        createNewSubs :: [RcvQueue] -> AM ()
+        createNewSubs rqs = do
           withTokenServer $ \ntfServer -> do
-            case clientNtfCreds of
-              Just ClientNtfCreds {notifierId} -> do
-                let newSub = newNtfSubscription connId smpServer (Just notifierId) ntfServer NASKey
-                withStore c $ \db -> createNtfSubscription db newSub $ NtfSubNTFAction NSACreate
-                lift . void $ getNtfNTFWorker True c ntfServer
-              Nothing -> do
-                let newSub = newNtfSubscription connId smpServer Nothing ntfServer NASNew
-                withStore c $ \db -> createNtfSubscription db newSub $ NtfSubSMPAction NSASmpKey
-                lift . void $ getNtfSMPWorker True c smpServer
-        (Just (sub@NtfSubscription {ntfSubStatus, ntfServer = subNtfServer, smpServer = smpServer', ntfQueueId}, action_)) -> do
-          case (clientNtfCreds, ntfQueueId) of
-            (Just ClientNtfCreds {notifierId}, Just ntfQueueId')
-              | sameSrvAddr smpServer smpServer' && notifierId == ntfQueueId' -> create
-              | otherwise -> rotate
-            (Nothing, Nothing) -> create
-            _ -> rotate
+            let newSubs = map (rqToNewSub ntfServer) rqs
+            (cErrs, _) <- lift $ partitionErrs ntfSubConnId newSubs <$> withStoreBatch c (\db -> map (storeNewSub db) newSubs)
+            notifyErrs c cErrs
+            kickSMPWorkers rqs
           where
-            create :: AM ()
-            create = case action_ of
-              -- action was set to NULL after worker internal error
-              Nothing -> resetSubscription
-              Just (action, _)
-                -- subscription was marked for deletion / is being deleted
-                | isDeleteNtfSubAction action -> do
-                    if ntfSubStatus == NASNew || ntfSubStatus == NASOff || ntfSubStatus == NASDeleted
-                      then resetSubscription
-                      else withTokenServer $ \ntfServer -> do
-                        withStore' c $ \db -> supervisorUpdateNtfSub db sub {ntfServer} (NtfSubNTFAction NSACreate)
-                        lift . void $ getNtfNTFWorker True c ntfServer
-                | otherwise -> case action of
-                    NtfSubNTFAction _ -> lift . void $ getNtfNTFWorker True c subNtfServer
-                    NtfSubSMPAction _ -> lift . void $ getNtfSMPWorker True c smpServer
-            rotate :: AM ()
-            rotate = do
-              withStore' c $ \db -> supervisorUpdateNtfSub db sub (NtfSubNTFAction NSARotate)
-              lift . void $ getNtfNTFWorker True c subNtfServer
-            resetSubscription :: AM ()
-            resetSubscription =
-              withTokenServer $ \ntfServer -> do
-                let sub' = sub {ntfQueueId = Nothing, ntfServer, ntfSubId = Nothing, ntfSubStatus = NASNew}
-                withStore' c $ \db -> supervisorUpdateNtfSub db sub' (NtfSubSMPAction NSASmpKey)
-                lift . void $ getNtfSMPWorker True c smpServer
-    NSCDelete -> do
-      sub_ <- withStore' c $ \db -> do
-        supervisorUpdateNtfAction db connId (NtfSubNTFAction NSADelete)
-        getNtfSubscription db connId
-      logInfo $ "processNtfSub, NSCDelete - sub_ = " <> tshow sub_
-      case sub_ of
-        (Just (NtfSubscription {ntfServer}, _)) -> lift . void $ getNtfNTFWorker True c ntfServer
-        _ -> pure () -- err "NSCDelete - no subscription"
+            rqToNewSub :: NtfServer -> RcvQueue -> NtfSubscription
+            rqToNewSub ntfServer RcvQueue {userId, connId, server} = newNtfSubscription userId connId server Nothing ntfServer NASNew
+            storeNewSub :: DB.Connection -> NtfSubscription -> IO (Either AgentErrorType ())
+            storeNewSub db sub = first storeError <$> createNtfSubscription db sub (NSASMP NSASmpKey)
+        resetSubs :: [(RcvQueue, NtfSubscription)] -> AM ()
+        resetSubs rqSubs = do
+          withTokenServer $ \ntfServer -> do
+            let subsToReset = map (toResetSub ntfServer) rqSubs
+            (cErrs, _) <- lift $ partitionErrs ntfSubConnId subsToReset <$> withStoreBatch' c (\db -> map (storeResetSub db) subsToReset)
+            notifyErrs c cErrs
+            let rqs = map fst rqSubs
+            kickSMPWorkers rqs
+          where
+            toResetSub :: NtfServer -> (RcvQueue, NtfSubscription) -> NtfSubscription
+            toResetSub ntfServer (rq, sub) =
+              let RcvQueue {server = smpServer} = rq
+               in sub {smpServer, ntfQueueId = Nothing, ntfServer, ntfSubId = Nothing, ntfSubStatus = NASNew}
+            storeResetSub :: DB.Connection -> NtfSubscription -> IO ()
+            storeResetSub db sub = supervisorUpdateNtfSub db sub (NSASMP NSASmpKey)
+        partitionQueueSubActions ::
+          [(RcvQueue, Maybe NtfSupervisorSub)] ->
+          ( [RcvQueue], -- new subs
+            [(RcvQueue, NtfSubscription)], -- reset subs
+            [SMPServer], -- continue work (SMP)
+            [NtfServer] -- continue work (Ntf)
+          )
+        partitionQueueSubActions = foldr decideSubWork ([], [], [], [])
+          where
+            -- sub = Nothing, needs to be created
+            decideSubWork (rq, Nothing) (ns, rs, css, cns) = (rq : ns, rs, css, cns)
+            decideSubWork (rq, Just (sub, subAction_)) (ns, rs, css, cns) =
+              case (clientNtfCreds rq, ntfQueueId sub) of
+                -- notifier ID created on SMP server (on ntf server subscription can be registered or not yet),
+                -- need to clarify action
+                (Just ClientNtfCreds {notifierId}, Just ntfQueueId')
+                  | sameSrvAddr (qServer rq) subSMPServer && notifierId == ntfQueueId' -> contOrReset
+                  | otherwise -> reset
+                (Nothing, Nothing) -> contOrReset
+                _ -> reset
+              where
+                NtfSubscription {ntfServer = subNtfServer, smpServer = subSMPServer} = sub
+                contOrReset = case subAction_ of
+                  -- action was set to NULL after worker internal error
+                  Nothing -> reset
+                  Just (action, _)
+                    -- subscription was marked for deletion / is being deleted
+                    | isDeleteNtfSubAction action -> reset
+                    -- continue work on subscription (e.g. supervisor was repeatedly tasked with creating a subscription)
+                    | otherwise -> case action of
+                        NSASMP _ -> (ns, rs, qServer rq : css, cns)
+                        NSANtf _ -> (ns, rs, css, subNtfServer : cns)
+                reset = (ns, (rq, sub) : rs, css, cns)
     NSCSmpDelete -> do
-      withStore' c (`getPrimaryRcvQueue` connId) >>= \case
-        Right rq@RcvQueue {server = smpServer} -> do
-          logInfo $ "processNtfSub, NSCSmpDelete - rq = " <> tshow rq
-          withStore' c $ \db -> supervisorUpdateNtfAction db connId (NtfSubSMPAction NSASmpDelete)
-          lift . void $ getNtfSMPWorker True c smpServer
-        _ -> notifyInternalError c connId "NSCSmpDelete - no rcv queue"
-    NSCNtfWorker ntfServer -> lift . void $ getNtfNTFWorker True c ntfServer
-    NSCNtfSMPWorker smpServer -> lift . void $ getNtfSMPWorker True c smpServer
+      (cErrs, rqs) <- lift $ partitionErrs id connIds' <$> withStoreBatch c (\db -> map (getQueue db) connIds')
+      logInfo $ "processNtfCmd, NSCSmpDelete - length rqs = " <> tshow (length rqs)
+      (cErrs', _) <- lift $ partitionErrs qConnId rqs <$> withStoreBatch' c (\db -> map (updateAction db) rqs)
+      notifyErrs c (cErrs <> cErrs')
+      kickSMPWorkers rqs
+      where
+        getQueue :: DB.Connection -> ConnId -> IO (Either AgentErrorType RcvQueue)
+        getQueue db connId = first storeError <$> getPrimaryRcvQueue db connId
+        updateAction :: DB.Connection -> RcvQueue -> IO ()
+        updateAction db rq = supervisorUpdateNtfAction db (qConnId rq) (NSASMP NSASmpDelete)
+    NSCDeleteSub -> void $ lift $ withStoreBatch' c $ \db -> map (deleteNtfSubscription' db) connIds'
+  where
+    kickSMPWorkers :: [RcvQueue] -> AM ()
+    kickSMPWorkers rqs = do
+      let smpServers = S.fromList $ map qServer rqs
+      lift $ mapM_ (getNtfSMPWorker True c) smpServers
 
 getNtfNTFWorker :: Bool -> AgentClient -> NtfServer -> AM' Worker
 getNtfNTFWorker hasWork c server = do
@@ -142,162 +198,303 @@ getNtfSMPWorker hasWork c server = do
   ws <- asks $ ntfSMPWorkers . ntfSupervisor
   getAgentWorker "ntf_smp" hasWork c server ws $ runNtfSMPWorker c server
 
+getNtfTknDelWorker :: Bool -> AgentClient -> NtfServer -> AM' Worker
+getNtfTknDelWorker hasWork c server = do
+  ws <- asks $ ntfTknDelWorkers . ntfSupervisor
+  getAgentWorker "ntf_tkn_del" hasWork c server ws $ runNtfTknDelWorker c server
+
 withTokenServer :: (NtfServer -> AM ()) -> AM ()
 withTokenServer action = lift getNtfToken >>= mapM_ (\NtfToken {ntfServer} -> action ntfServer)
 
 runNtfWorker :: AgentClient -> NtfServer -> Worker -> AM ()
-runNtfWorker c srv Worker {doWork} = do
-  delay <- asks $ ntfWorkerDelay . config
+runNtfWorker c srv Worker {doWork} =
   forever $ do
     waitForWork doWork
     ExceptT $ agentOperationBracket c AONtfNetwork throwWhenInactive $ runExceptT runNtfOperation
-    threadDelay delay
   where
     runNtfOperation :: AM ()
-    runNtfOperation =
-      withWork c doWork (`getNextNtfSubNTFAction` srv) $
-        \nextSub@(NtfSubscription {connId}, _, _) -> do
-          logInfo $ "runNtfWorker, nextSub " <> tshow nextSub
-          ri <- asks $ reconnectInterval . config
-          withRetryInterval ri $ \_ loop -> do
-            liftIO $ waitForUserNetwork c
-            processSub nextSub
-              `catchAgentError` retryOnError c "NtfWorker" loop (workerInternalError c connId . show)
-    processSub :: (NtfSubscription, NtfSubNTFAction, NtfActionTs) -> AM ()
-    processSub (sub@NtfSubscription {connId, smpServer, ntfSubId}, action, actionTs) = do
-      ts <- liftIO getCurrentTime
-      unlessM (lift $ rescheduleAction doWork ts actionTs) $
-        case action of
-          NSACreate ->
-            lift getNtfToken >>= \case
-              Just tkn@NtfToken {ntfTokenId = Just tknId, ntfTknStatus = NTActive, ntfMode = NMInstant} -> do
-                RcvQueue {clientNtfCreds} <- withStore c (`getPrimaryRcvQueue` connId)
-                case clientNtfCreds of
-                  Just ClientNtfCreds {ntfPrivateKey, notifierId} -> do
-                    nSubId <- agentNtfCreateSubscription c tknId tkn (SMPQueueNtf smpServer notifierId) ntfPrivateKey
-                    -- possible improvement: smaller retry until Active, less frequently (daily?) once Active
-                    let actionTs' = addUTCTime 30 ts
-                    withStore' c $ \db ->
-                      updateNtfSubscription db sub {ntfSubId = Just nSubId, ntfSubStatus = NASCreated NSNew} (NtfSubNTFAction NSACheck) actionTs'
-                  _ -> workerInternalError c connId "NSACreate - no notifier queue credentials"
-              _ -> workerInternalError c connId "NSACreate - no active token"
-          NSACheck ->
-            lift getNtfToken >>= \case
-              Just tkn ->
-                case ntfSubId of
-                  Just nSubId ->
-                    agentNtfCheckSubscription c nSubId tkn >>= \case
-                      NSAuth -> do
-                        lift (getNtfServer c) >>= \case
-                          Just ntfServer -> do
-                            withStore' c $ \db ->
-                              updateNtfSubscription db sub {ntfServer, ntfQueueId = Nothing, ntfSubId = Nothing, ntfSubStatus = NASNew} (NtfSubSMPAction NSASmpKey) ts
-                            ns <- asks ntfSupervisor
-                            atomically $ writeTBQueue (ntfSubQ ns) (connId, NSCNtfSMPWorker smpServer)
-                          _ -> workerInternalError c connId "NSACheck - failed to reset subscription, notification server not configured"
-                      status -> updateSubNextCheck ts status
-                  Nothing -> workerInternalError c connId "NSACheck - no subscription ID"
-              _ -> workerInternalError c connId "NSACheck - no active token"
-          NSADelete -> case ntfSubId of
-            Just nSubId ->
-              (lift getNtfToken >>= mapM_ (agentNtfDeleteSubscription c nSubId))
-                `agentFinally` continueDeletion
-            _ -> continueDeletion
-            where
-              continueDeletion = do
-                let sub' = sub {ntfSubId = Nothing, ntfSubStatus = NASOff}
-                withStore' c $ \db -> updateNtfSubscription db sub' (NtfSubSMPAction NSASmpDelete) ts
-                ns <- asks ntfSupervisor
-                atomically $ writeTBQueue (ntfSubQ ns) (connId, NSCNtfSMPWorker smpServer)
-          NSARotate -> case ntfSubId of
-            Just nSubId ->
-              (lift getNtfToken >>= mapM_ (agentNtfDeleteSubscription c nSubId))
-                `agentFinally` deleteCreate
-            _ -> deleteCreate
-            where
-              deleteCreate = do
-                withStore' c $ \db -> deleteNtfSubscription db connId
-                ns <- asks ntfSupervisor
-                atomically $ writeTBQueue (ntfSubQ ns) (connId, NSCCreate)
+    runNtfOperation = do
+      ntfBatchSize <- asks $ ntfBatchSize . config
+      withWorkItems c doWork (\db -> getNextNtfSubNTFActions db srv ntfBatchSize) $ \nextSubs -> do
+        logInfo $ "runNtfWorker - length nextSubs = " <> tshow (length nextSubs)
+        currTs <- liftIO getCurrentTime
+        let (creates, checks, deletes, rotates) = splitActions currTs nextSubs
+        if null creates && null checks && null deletes && null rotates
+          then
+            let (_, _, firstActionTs) = L.head nextSubs
+             in lift $ rescheduleWork doWork currTs firstActionTs
+          else do
+            retrySubActions c creates createSubs
+            retrySubActions c checks checkSubs
+            retrySubActions c deletes deleteSubs
+            retrySubActions c rotates rotateSubs
+    splitActions :: UTCTime -> NonEmpty (NtfSubNTFAction, NtfSubscription, NtfActionTs) -> ([NtfSubscription], [NtfSubscription], [NtfSubscription], [NtfSubscription])
+    splitActions currTs = foldr addAction ([], [], [], [])
       where
-        updateSubNextCheck ts toStatus = do
-          checkInterval <- asks $ ntfSubCheckInterval . config
-          let nextCheckTs = addUTCTime checkInterval ts
-          updateSub (NASCreated toStatus) (NtfSubNTFAction NSACheck) nextCheckTs
-        updateSub toStatus toAction actionTs' =
-          withStore' c $ \db ->
-            updateNtfSubscription db sub {ntfSubStatus = toStatus} toAction actionTs'
+        addAction (cmd, sub, ts) acc@(creates, checks, deletes, rotates) = case cmd of
+          NSACreate -> (sub : creates, checks, deletes, rotates)
+          NSACheck
+            | ts <= currTs -> (creates, sub : checks, deletes, rotates)
+            | otherwise -> acc
+          NSADelete -> (creates, checks, sub : deletes, rotates)
+          NSARotate -> (creates, checks, deletes, sub : rotates)
+    createSubs :: [NtfSubscription] -> AM' [NtfSubscription]
+    createSubs ntfSubs =
+      getNtfToken >>= \case
+        Just tkn@NtfToken {ntfServer, ntfTokenId = Just tknId, ntfTknStatus = NTActive, ntfMode = NMInstant} -> do
+          subsRqs_ <- zip ntfSubs <$> withStoreBatch c (\db -> map (getQueue db) ntfSubs)
+          let (errs1, subs_, newSubs_) = splitSubs tknId subsRqs_
+          incStatByUserId ntfServer ntfCreateAttempts subs_
+          case (L.nonEmpty subs_, L.nonEmpty newSubs_) of
+            (Just subs, Just newSubs) -> do
+              rs <- L.zip subs <$> agentNtfCreateSubscriptions c tkn newSubs
+              let (ntfSubs', errs2, nSubIds) = splitResults $ L.toList rs
+                  subs' = map fst nSubIds
+                  errs2' = map (first ntfSubConnId) errs2
+              incStatByUserId ntfServer ntfCreated subs'
+              ts <- liftIO getCurrentTime
+              int <- asks $ ntfSubFirstCheckInterval . config
+              let checkTs = addUTCTime int ts
+              (errs3, _) <- partitionErrs ntfSubConnId subs' <$> withStoreBatch' c (\db -> map (updateSubNSACheck db checkTs) nSubIds)
+              workerErrors c $ errs1 <> errs2' <> errs3
+              pure ntfSubs'
+            _ -> workerErrors c errs1 $> []
+        _ -> do
+          let errs = map (\sub -> (ntfSubConnId sub, INTERNAL "NSACreate - no active token")) ntfSubs
+          workerErrors c errs
+          pure []
+      where
+        getQueue :: DB.Connection -> NtfSubscription -> IO (Either AgentErrorType RcvQueue)
+        getQueue db NtfSubscription {connId} = first storeError <$> getPrimaryRcvQueue db connId
+        splitSubs :: NtfTokenId -> [(NtfSubscription, Either AgentErrorType RcvQueue)] -> ([(ConnId, AgentErrorType)], [NtfSubscription], [NewNtfEntity 'Subscription])
+        splitSubs tknId = foldr splitSub ([], [], [])
+          where
+            splitSub (sub, rq) (errs, subs, newSubs) = case rq of
+              Right RcvQueue {clientNtfCreds = Just creds} -> (errs, sub : subs, toNewSub sub creds : newSubs)
+              Right _ -> ((ntfSubConnId sub, INTERNAL "NSACreate - no notifier queue credentials") : errs, subs, newSubs)
+              Left e -> ((ntfSubConnId sub, e) : errs, subs, newSubs)
+            toNewSub NtfSubscription {smpServer} ClientNtfCreds {ntfPrivateKey, notifierId} =
+              NewNtfSub tknId (SMPQueueNtf smpServer notifierId) ntfPrivateKey
+        updateSubNSACheck :: DB.Connection -> UTCTime -> (NtfSubscription, NtfSubscriptionId) -> IO ()
+        updateSubNSACheck db checkTs (sub, nSubId) = updateNtfSubscription db sub {ntfSubId = Just nSubId, ntfSubStatus = NASCreated NSNew} (NSANtf NSACheck) checkTs
+    checkSubs :: [NtfSubscription] -> AM' [NtfSubscription]
+    checkSubs ntfSubs =
+      getNtfToken >>= \case
+        Just tkn@NtfToken {ntfServer, ntfTknStatus = NTActive, ntfMode = NMInstant} -> do
+          let (errs1, subs_, subIds_) = splitSubs ntfSubs
+          incStatByUserId ntfServer ntfCheckAttempts subs_
+          case (L.nonEmpty subs_, L.nonEmpty subIds_) of
+            (Just subs, Just subIds) -> do
+              rs <- L.zip subs <$> agentNtfCheckSubscriptions c tkn subIds
+              let (ntfSubs', errs2, nSubStatuses) = splitResults $ L.toList rs
+                  subs' = map fst nSubStatuses
+                  (errs2', authSubs) = partitionEithers $ map (\case (sub, NTF _ SMP.AUTH) -> Right sub; e -> Left $ first ntfSubConnId e) errs2
+              incStatByUserId ntfServer ntfChecked subs'
+              ts <- liftIO getCurrentTime
+              int <- asks $ ntfSubCheckInterval . config
+              let nextCheckTs = addUTCTime int ts
+              (errs3, srvs) <- partitionErrs ntfSubConnId subs' <$> withStoreBatch' c (\db -> map (updateSub db ntfServer ts nextCheckTs) nSubStatuses)
+              (errs4, srvs') <- partitionErrs ntfSubConnId authSubs <$> withStoreBatch' c (\db -> map (recreateNtfSub db ntfServer ts) authSubs)
+              mapM_ (getNtfSMPWorker True c) $ S.fromList (catMaybes srvs <> srvs')
+              workerErrors c $ errs1 <> errs2' <> errs3 <> errs4
+              pure ntfSubs'
+            _ -> workerErrors c errs1 $> []
+        _ -> do
+          let errs = map (\sub -> (ntfSubConnId sub, INTERNAL "NSACheck - no active token")) ntfSubs
+          workerErrors c errs
+          pure []
+      where
+        splitSubs :: [NtfSubscription] -> ([(ConnId, AgentErrorType)], [NtfSubscription], [NtfSubscriptionId])
+        splitSubs = foldr splitSub ([], [], [])
+          where
+            splitSub sub (errs, subs, subIds) = case sub of
+              NtfSubscription {ntfSubId = Just subId} -> (errs, sub : subs, subId : subIds)
+              _ -> ((ntfSubConnId sub, INTERNAL "NSACheck - no subscription ID") : errs, subs, subIds)
+        updateSub :: DB.Connection -> NtfServer -> UTCTime -> UTCTime -> (NtfSubscription, NtfSubStatus) -> IO (Maybe SMPServer)
+        updateSub db ntfServer ts nextCheckTs (sub, status)
+          | ntfShouldSubscribe status =
+              let sub' = sub {ntfSubStatus = NASCreated status}
+               in Nothing <$ updateNtfSubscription db sub' (NSANtf NSACheck) nextCheckTs
+          -- ntf server stopped subscribing to this queue
+          | otherwise = Just <$> recreateNtfSub db ntfServer ts sub
+        recreateNtfSub :: DB.Connection -> NtfServer -> UTCTime -> NtfSubscription -> IO SMPServer
+        recreateNtfSub db ntfServer ts sub@NtfSubscription {smpServer} =
+          let sub' = sub {ntfServer, ntfQueueId = Nothing, ntfSubId = Nothing, ntfSubStatus = NASNew}
+           in smpServer <$ updateNtfSubscription db sub' (NSASMP NSASmpKey) ts
+    incStatByUserId :: NtfServer -> (AgentNtfServerStats -> TVar Int) -> [NtfSubscription] -> AM' ()
+    incStatByUserId ntfServer sel ss =
+      forM_ (M.assocs userIdsCounts) $ \(userId, count) ->
+        atomically $ incNtfServerStat' c userId ntfServer sel count
+      where
+        userIdsCounts = foldl' (\acc NtfSubscription {userId} -> M.insertWith (+) userId 1 acc) M.empty ss
+    -- NSADelete and NSARotate are deprecated, but their processing is kept for legacy db records;
+    -- These actions are not batched
+    deleteSubs :: [NtfSubscription] -> AM' [NtfSubscription]
+    deleteSubs ntfSubs = do
+      retrySubs_ <- mapM (runCatching deleteSub) ntfSubs
+      pure $ catMaybes retrySubs_
+      where
+        deleteSub :: NtfSubscription -> AM (Maybe NtfSubscription)
+        deleteSub sub@NtfSubscription {smpServer} =
+          deleteNtfSub sub $ do
+            let sub' = sub {ntfSubId = Nothing, ntfSubStatus = NASOff}
+            ts <- liftIO getCurrentTime
+            withStore' c $ \db -> updateNtfSubscription db sub' (NSASMP NSASmpDelete) ts
+            lift . void $ getNtfSMPWorker True c smpServer
+    rotateSubs :: [NtfSubscription] -> AM' [NtfSubscription]
+    rotateSubs ntfSubs = do
+      retrySubs_ <- mapM (runCatching rotateSub) ntfSubs
+      pure $ catMaybes retrySubs_
+      where
+        rotateSub :: NtfSubscription -> AM (Maybe NtfSubscription)
+        rotateSub sub@NtfSubscription {connId} =
+          deleteNtfSub sub $ do
+            withStore' c $ \db -> deleteNtfSubscription db connId
+            ns <- asks ntfSupervisor
+            atomically $ writeTBQueue (ntfSubQ ns) (NSCCreate, [connId])
+    runCatching :: (NtfSubscription -> AM (Maybe NtfSubscription)) -> NtfSubscription -> AM' (Maybe NtfSubscription)
+    runCatching action sub@NtfSubscription {connId} =
+      fromRight Nothing
+        <$> runExceptT (action sub `catchAgentError` \e -> workerInternalError c connId (show e) $> Nothing)
+    -- deleteNtfSub is only used in NSADelete and NSARotate, so also deprecated
+    deleteNtfSub :: NtfSubscription -> AM () -> AM (Maybe NtfSubscription)
+    deleteNtfSub sub@NtfSubscription {userId, ntfSubId} continue = case ntfSubId of
+      Just nSubId ->
+        lift getNtfToken >>= \case
+          Just tkn@NtfToken {ntfServer} -> do
+            atomically $ incNtfServerStat c userId ntfServer ntfDelAttempts
+            tryAgentError (agentNtfDeleteSubscription c nSubId tkn) >>= \case
+              Right _ -> do
+                atomically $ incNtfServerStat c userId ntfServer ntfDeleted
+                continue'
+              Left e
+                | temporaryOrHostError e -> pure $ Just sub -- don't continue, retry
+                | otherwise -> continue'
+          Nothing -> continue'
+      _ -> continue'
+      where
+        continue' = continue $> Nothing -- continue without retry
 
 runNtfSMPWorker :: AgentClient -> SMPServer -> Worker -> AM ()
-runNtfSMPWorker c srv Worker {doWork} = do
-  env <- ask
-  delay <- asks $ ntfSMPWorkerDelay . config
-  forever $ do
-    waitForWork doWork
-    ExceptT . liftIO . agentOperationBracket c AONtfNetwork throwWhenInactive $
-      runReaderT (runExceptT runNtfSMPOperation) env
-    threadDelay delay
+runNtfSMPWorker c srv Worker {doWork} = forever $ do
+  waitForWork doWork
+  ExceptT $ agentOperationBracket c AONtfNetwork throwWhenInactive $ runExceptT runNtfSMPOperation
   where
-    runNtfSMPOperation =
-      withWork c doWork (`getNextNtfSubSMPAction` srv) $
-        \nextSub@(NtfSubscription {connId}, _, _) -> do
-          logInfo $ "runNtfSMPWorker, nextSub " <> tshow nextSub
-          ri <- asks $ reconnectInterval . config
-          withRetryInterval ri $ \_ loop -> do
-            liftIO $ waitForUserNetwork c
-            processSub nextSub
-              `catchAgentError` retryOnError c "NtfSMPWorker" loop (workerInternalError c connId . show)
-    processSub :: (NtfSubscription, NtfSubSMPAction, NtfActionTs) -> AM ()
-    processSub (sub@NtfSubscription {connId, ntfServer}, smpAction, actionTs) = do
-      ts <- liftIO getCurrentTime
-      unlessM (lift $ rescheduleAction doWork ts actionTs) $
-        case smpAction of
-          NSASmpKey ->
-            lift getNtfToken >>= \case
-              Just NtfToken {ntfTknStatus = NTActive, ntfMode = NMInstant} -> do
-                rq <- withStore c (`getPrimaryRcvQueue` connId)
-                C.AuthAlg a <- asks (rcvAuthAlg . config)
-                g <- asks random
-                (ntfPublicKey, ntfPrivateKey) <- atomically $ C.generateAuthKeyPair a g
-                (rcvNtfPubDhKey, rcvNtfPrivDhKey) <- atomically $ C.generateKeyPair g
-                (notifierId, rcvNtfSrvPubDhKey) <- enableQueueNotifications c rq ntfPublicKey rcvNtfPubDhKey
-                let rcvNtfDhSecret = C.dh' rcvNtfSrvPubDhKey rcvNtfPrivDhKey
-                withStore' c $ \db -> do
-                  setRcvQueueNtfCreds db connId $ Just ClientNtfCreds {ntfPublicKey, ntfPrivateKey, notifierId, rcvNtfDhSecret}
-                  updateNtfSubscription db sub {ntfQueueId = Just notifierId, ntfSubStatus = NASKey} (NtfSubNTFAction NSACreate) ts
-                ns <- asks ntfSupervisor
-                atomically $ sendNtfSubCommand ns (connId, NSCNtfWorker ntfServer)
-              _ -> workerInternalError c connId "NSASmpKey - no active token"
-          NSASmpDelete -> do
-            rq_ <- withStore' c $ \db -> do
-              setRcvQueueNtfCreds db connId Nothing
-              getPrimaryRcvQueue db connId
-            mapM_ (disableQueueNotifications c) rq_
-            withStore' c $ \db -> deleteNtfSubscription db connId
+    runNtfSMPOperation :: AM ()
+    runNtfSMPOperation = do
+      ntfBatchSize <- asks $ ntfBatchSize . config
+      withWorkItems c doWork (\db -> getNextNtfSubSMPActions db srv ntfBatchSize) $ \nextSubs -> do
+        logInfo $ "runNtfSMPWorker - length nextSubs = " <> tshow (length nextSubs)
+        let (creates, deletes) = splitActions nextSubs
+        retrySubActions c creates createNotifierKeys
+        retrySubActions c deletes deleteNotifierKeys
+    splitActions :: NonEmpty (NtfSubSMPAction, NtfSubscription) -> ([NtfSubscription], [NtfSubscription])
+    splitActions = foldr addAction ([], [])
+      where
+        addAction (cmd, sub) (creates, deletes) = case cmd of
+          NSASmpKey -> (sub : creates, deletes)
+          NSASmpDelete -> (creates, sub : deletes)
+    createNotifierKeys :: [NtfSubscription] -> AM' [NtfSubscription]
+    createNotifierKeys ntfSubs =
+      getNtfToken >>= \case
+        Just NtfToken {ntfTknStatus = NTActive, ntfMode = NMInstant} -> do
+          (errs1, subRqKeys) <- prepareQueueSmpKey ntfSubs
+          rs <- enableQueuesNtfs c subRqKeys
+          let (subRqKeys', errs2, successes) = splitResults rs
+              ntfSubs' = map eqnrNtfSub subRqKeys'
+              errs2' = map (first (qConnId . eqnrRq)) errs2
+          ts <- liftIO getCurrentTime
+          (errs3, srvs) <- partitionErrs (qConnId . eqnrRq . fst) successes <$> withStoreBatch' c (\db -> map (storeNtfSubCreds db ts) successes)
+          mapM_ (getNtfNTFWorker True c) $ S.fromList srvs
+          workerErrors c $ errs1 <> errs2' <> errs3
+          pure ntfSubs'
+        _ -> do
+          let errs = map (\sub -> (ntfSubConnId sub, INTERNAL "NSASmpKey - no active token")) ntfSubs
+          workerErrors c errs
+          pure []
+      where
+        prepareQueueSmpKey :: [NtfSubscription] -> AM' ([(ConnId, AgentErrorType)], [EnableQueueNtfReq])
+        prepareQueueSmpKey subs = do
+          alg <- asks (rcvAuthAlg . config)
+          g <- asks random
+          partitionErrs ntfSubConnId subs <$> withStoreBatch c (\db -> map (getQueue db alg g) subs)
+          where
+            getQueue :: DB.Connection -> C.AuthAlg -> TVar ChaChaDRG -> NtfSubscription -> IO (Either AgentErrorType EnableQueueNtfReq)
+            getQueue db (C.AuthAlg a) g sub = fmap (first storeError) $ runExceptT $ do
+              rq <- ExceptT $ getPrimaryRcvQueue db (ntfSubConnId sub)
+              authKeyPair <- atomically $ C.generateAuthKeyPair a g
+              rcvNtfKeyPair <- atomically $ C.generateKeyPair g
+              pure (EnableQueueNtfReq sub rq authKeyPair rcvNtfKeyPair)
+        storeNtfSubCreds :: DB.Connection -> UTCTime -> (EnableQueueNtfReq, (SMP.NotifierId, SMP.RcvNtfPublicDhKey)) -> IO NtfServer
+        storeNtfSubCreds db ts (EnableQueueNtfReq {eqnrNtfSub, eqnrAuthKeyPair = (ntfPublicKey, ntfPrivateKey), eqnrRcvKeyPair = (_, pk)}, (notifierId, srvPubDhKey)) = do
+          let NtfSubscription {ntfServer} = eqnrNtfSub
+              rcvNtfDhSecret = C.dh' srvPubDhKey pk
+          setRcvQueueNtfCreds db (ntfSubConnId eqnrNtfSub) $ Just ClientNtfCreds {ntfPublicKey, ntfPrivateKey, notifierId, rcvNtfDhSecret}
+          updateNtfSubscription db eqnrNtfSub {ntfQueueId = Just notifierId, ntfSubStatus = NASKey} (NSANtf NSACreate) ts
+          pure ntfServer
+    deleteNotifierKeys :: [NtfSubscription] -> AM' [NtfSubscription]
+    deleteNotifierKeys ntfSubs = do
+      (errs1, subRqs) <- partitionErrs ntfSubConnId ntfSubs <$> withStoreBatch c (\db -> map (resetCredsGetQueue db) ntfSubs)
+      rs <- disableQueuesNtfs c subRqs
+      let (subRqs', errs2, successes) = splitResults rs
+          ntfSubs' = map fst subRqs'
+          errs2' = map (first (qConnId . snd)) errs2
+          disabledRqs = map (snd . fst) successes
+      (errs3, _) <- partitionErrs qConnId disabledRqs <$> withStoreBatch' c (\db -> map (deleteSub db) disabledRqs)
+      workerErrors c $ errs1 <> errs2' <> errs3
+      pure ntfSubs'
+      where
+        resetCredsGetQueue :: DB.Connection -> NtfSubscription -> IO (Either AgentErrorType DisableQueueNtfReq)
+        resetCredsGetQueue db sub@NtfSubscription {connId} = fmap (first storeError) $ runExceptT $ do
+          liftIO $ setRcvQueueNtfCreds db connId Nothing
+          rq <- ExceptT $ getPrimaryRcvQueue db connId
+          pure (sub, rq)
+        deleteSub :: DB.Connection -> RcvQueue -> IO ()
+        deleteSub db rq = deleteNtfSubscription db (qConnId rq)
 
-rescheduleAction :: TMVar () -> UTCTime -> UTCTime -> AM' Bool
-rescheduleAction doWork ts actionTs
-  | actionTs <= ts = pure False
-  | otherwise = do
-      void . atomically $ tryTakeTMVar doWork
-      void . forkIO $ do
-        liftIO $ threadDelay' $ diffToMicroseconds $ diffUTCTime actionTs ts
-        atomically $ hasWorkToDo' doWork
-      pure True
+retrySubActions :: AgentClient -> [NtfSubscription] -> ([NtfSubscription] -> AM' [NtfSubscription]) -> AM ()
+retrySubActions _ [] _ = pure ()
+retrySubActions c subs action = do
+  v <- newTVarIO subs
+  ri <- asks $ reconnectInterval . config
+  withRetryInterval ri $ \_ loop -> do
+    liftIO $ waitWhileSuspended c
+    liftIO $ waitForUserNetwork c
+    subs' <- readTVarIO v
+    retrySubs <- lift $ action subs'
+    unless (null retrySubs) $ do
+      atomically $ writeTVar v retrySubs
+      retryNetworkLoop c loop
 
-retryOnError :: AgentClient -> Text -> AM () -> (AgentErrorType -> AM ()) -> AgentErrorType -> AM ()
-retryOnError c name loop done e = do
-  logError $ name <> " error: " <> tshow e
-  case e of
-    BROKER _ NETWORK -> retryLoop
-    BROKER _ TIMEOUT -> retryLoop
-    _ -> done e
+--                                                (temporary errs, other errs, successes)
+splitResults :: [(a, Either AgentErrorType r)] -> ([a], [(a, AgentErrorType)], [(a, r)])
+splitResults = foldr addRes ([], [], [])
   where
-    retryLoop = do
-      atomically $ endAgentOperation c AONtfNetwork
-      atomically $ throwWhenInactive c
-      atomically $ beginAgentOperation c AONtfNetwork
-      loop
+    addRes (a, r_) (as, errs, rs) = case r_ of
+      Right r -> (as, errs, (a, r) : rs)
+      Left e
+        | temporaryOrHostError e -> (a : as, errs, rs)
+        | otherwise -> (as, (a, e) : errs, rs)
+
+rescheduleWork :: TMVar () -> UTCTime -> UTCTime -> AM' ()
+rescheduleWork doWork ts actionTs = do
+  void . atomically $ tryTakeTMVar doWork
+  void . forkIO $ do
+    liftIO $ threadDelay' $ diffToMicroseconds $ diffUTCTime actionTs ts
+    atomically $ hasWorkToDo' doWork
+
+retryNetworkLoop :: AgentClient -> AM () -> AM ()
+retryNetworkLoop c loop = do
+  atomically $ endAgentOperation c AONtfNetwork
+  liftIO $ throwWhenInactive c
+  atomically $ beginAgentOperation c AONtfNetwork
+  loop
+
+workerErrors :: AgentClient -> [(ConnId, AgentErrorType)] -> AM' ()
+workerErrors c connErrs =
+  unless (null connErrs) $ do
+    void $ withStoreBatch' c (\db -> map (setNullNtfSubscriptionAction db . fst) connErrs)
+    notifyErrs c connErrs
 
 workerInternalError :: AgentClient -> ConnId -> String -> AM ()
 workerInternalError c connId internalErrStr = do
@@ -309,6 +506,14 @@ notifyInternalError :: MonadIO m => AgentClient -> ConnId -> String -> m ()
 notifyInternalError AgentClient {subQ} connId internalErrStr = atomically $ writeTBQueue subQ ("", connId, AEvt SAEConn $ ERR $ INTERNAL internalErrStr)
 {-# INLINE notifyInternalError #-}
 
+notifyInternalError' :: MonadIO m => AgentClient -> String -> m ()
+notifyInternalError' AgentClient {subQ} internalErrStr = atomically $ writeTBQueue subQ ("", "", AEvt SAEConn $ ERR $ INTERNAL internalErrStr)
+{-# INLINE notifyInternalError' #-}
+
+notifyErrs :: MonadIO m => AgentClient -> [(ConnId, AgentErrorType)] -> m ()
+notifyErrs AgentClient {subQ} connErrs = unless (null connErrs) $ atomically $ writeTBQueue subQ ("", "", AEvt SAENone $ ERRS connErrs)
+{-# INLINE notifyErrs #-}
+
 getNtfToken :: AM' (Maybe NtfToken)
 getNtfToken = do
   tkn <- asks $ ntfTkn . ntfSupervisor
@@ -320,7 +525,7 @@ nsUpdateToken ns tkn = writeTVar (ntfTkn ns) $ Just tkn
 nsRemoveNtfToken :: NtfSupervisor -> STM ()
 nsRemoveNtfToken ns = writeTVar (ntfTkn ns) Nothing
 
-sendNtfSubCommand :: NtfSupervisor -> (ConnId, NtfSupervisorCommand) -> STM ()
+sendNtfSubCommand :: NtfSupervisor -> (NtfSupervisorCommand, NonEmpty ConnId) -> STM ()
 sendNtfSubCommand ns cmd = do
   tkn <- readTVar (ntfTkn ns)
   when (instantNotifications tkn) $ writeTBQueue (ntfSubQ ns) cmd
@@ -330,10 +535,51 @@ instantNotifications = \case
   Just NtfToken {ntfTknStatus = NTActive, ntfMode = NMInstant} -> True
   _ -> False
 
+deleteToken :: AgentClient -> NtfToken -> AM ()
+deleteToken c tkn@NtfToken {ntfServer, ntfTokenId, ntfPrivKey} = do
+  setToDelete <- withStore' c $ \db -> do
+    removeNtfToken db tkn
+    case ntfTokenId of
+      Just tknId -> addNtfTokenToDelete db ntfServer ntfPrivKey tknId $> True
+      Nothing -> pure False
+  ns <- asks ntfSupervisor
+  atomically $ nsRemoveNtfToken ns
+  when setToDelete $ void $ lift $ getNtfTknDelWorker True c ntfServer
+
+runNtfTknDelWorker :: AgentClient -> NtfServer -> Worker -> AM ()
+runNtfTknDelWorker c srv Worker {doWork} =
+  forever $ do
+    waitForWork doWork
+    ExceptT $ agentOperationBracket c AONtfNetwork throwWhenInactive $ runExceptT runNtfOperation
+  where
+    runNtfOperation :: AM ()
+    runNtfOperation =
+      withWork c doWork (`getNextNtfTokenToDelete` srv) $
+        \nextTknToDelete -> do
+          logInfo $ "runNtfTknDelWorker, nextTknToDelete " <> tshow nextTknToDelete
+          ri <- asks $ reconnectInterval . config
+          withRetryInterval ri $ \_ loop -> do
+            liftIO $ waitWhileSuspended c
+            liftIO $ waitForUserNetwork c
+            processTknToDelete nextTknToDelete `catchAgentError` retryTmpError loop nextTknToDelete
+    retryTmpError :: AM () -> NtfTokenToDelete -> AgentErrorType -> AM ()
+    retryTmpError loop (tknDbId, _, _) e = do
+      logError $ "ntf tkn del error: " <> tshow e
+      if temporaryOrHostError e
+        then retryNetworkLoop c loop
+        else do
+          withStore' c $ \db -> deleteNtfTokenToDelete db tknDbId
+          notifyInternalError' c (show e)
+    processTknToDelete :: NtfTokenToDelete -> AM ()
+    processTknToDelete (tknDbId, ntfPrivKey, tknId) = do
+      agentNtfDeleteToken c srv ntfPrivKey tknId
+      withStore' c $ \db -> deleteNtfTokenToDelete db tknDbId
+
 closeNtfSupervisor :: NtfSupervisor -> IO ()
 closeNtfSupervisor ns = do
   stopWorkers $ ntfWorkers ns
   stopWorkers $ ntfSMPWorkers ns
+  stopWorkers $ ntfTknDelWorkers ns
   where
     stopWorkers workers = atomically (swapTVar workers M.empty) >>= mapM_ (liftIO . cancelWorker)
 

src/Simplex/Messaging/Agent/Protocol.hs

@@ -42,6 +42,7 @@ module Simplex.Messaging.Agent.Protocol
     deliveryRcptsSMPAgentVersion,
     pqdrSMPAgentVersion,
     sndAuthKeySMPAgentVersion,
+    ratchetOnConfSMPAgentVersion,
     currentSMPAgentVersion,
     supportedSMPAgentVRange,
     e2eEncConnInfoLength,
@@ -49,6 +50,8 @@ module Simplex.Messaging.Agent.Protocol
 
     -- * SMP agent protocol types
     ConnInfo,
+    SndQueueSecured,
+    AEntityId,
     ACommand (..),
     AEvent (..),
     AEvt (..),
@@ -105,6 +108,7 @@ module Simplex.Messaging.Agent.Protocol
     ConnReqUriData (..),
     CRClientData,
     ServiceScheme,
+    sameConnReqContact,
     simplexChat,
     connReqUriP',
     AgentErrorType (..),
@@ -153,8 +157,8 @@ import Data.Int (Int64)
 import Data.Kind (Type)
 import Data.List.NonEmpty (NonEmpty (..))
 import qualified Data.List.NonEmpty as L
-import Data.Map (Map)
-import qualified Data.Map as M
+import Data.Map.Strict (Map)
+import qualified Data.Map.Strict as M
 import Data.Maybe (fromMaybe, isJust)
 import Data.Text (Text)
 import Data.Text.Encoding (decodeLatin1, encodeUtf8)
@@ -188,7 +192,6 @@ import Simplex.Messaging.Parsers
 import Simplex.Messaging.Protocol
   ( AProtocolType,
     BrokerErrorType (..),
-    EntityId,
     ErrorType,
     MsgBody,
     MsgFlags,
@@ -196,7 +199,6 @@ import Simplex.Messaging.Protocol
     NMsgMeta,
     ProtocolServer (..),
     SMPClientVersion,
-    SMPMsgMeta,
     SMPServer,
     SMPServerWithAuth,
     SndPublicAuthKey,
@@ -257,11 +259,14 @@ pqdrSMPAgentVersion = VersionSMPA 5
 sndAuthKeySMPAgentVersion :: VersionSMPA
 sndAuthKeySMPAgentVersion = VersionSMPA 6
 
+ratchetOnConfSMPAgentVersion :: VersionSMPA
+ratchetOnConfSMPAgentVersion = VersionSMPA 7
+
 minSupportedSMPAgentVersion :: VersionSMPA
 minSupportedSMPAgentVersion = duplexHandshakeSMPAgentVersion
 
 currentSMPAgentVersion :: VersionSMPA
-currentSMPAgentVersion = VersionSMPA 6
+currentSMPAgentVersion = VersionSMPA 7
 
 supportedSMPAgentVRange :: VersionRangeSMPA
 supportedSMPAgentVRange = mkVersionRange minSupportedSMPAgentVersion currentSMPAgentVersion
@@ -272,20 +277,22 @@ supportedSMPAgentVRange = mkVersionRange minSupportedSMPAgentVersion currentSMPA
 e2eEncConnInfoLength :: VersionSMPA -> PQSupport -> Int
 e2eEncConnInfoLength v = \case
   -- reduced by 3726 (roughly the increase of message ratchet header size + key and ciphertext in reply link)
-  PQSupportOn | v >= pqdrSMPAgentVersion -> 11122
-  _ -> 14848
+  PQSupportOn | v >= pqdrSMPAgentVersion -> 11106
+  _ -> 14832
 
 e2eEncAgentMsgLength :: VersionSMPA -> PQSupport -> Int
 e2eEncAgentMsgLength v = \case
   -- reduced by 2222 (the increase of message ratchet header size)
-  PQSupportOn | v >= pqdrSMPAgentVersion -> 13634
-  _ -> 15856
+  PQSupportOn | v >= pqdrSMPAgentVersion -> 13618
+  _ -> 15840
 
 -- | SMP agent event
-type ATransmission = (ACorrId, EntityId, AEvt)
+type ATransmission = (ACorrId, AEntityId, AEvt)
 
 type UserId = Int64
 
+type AEntityId = ByteString
+
 type ACorrId = ByteString
 
 data AEntity = AEConn | AERcvFile | AESndFile | AENone
@@ -327,6 +334,8 @@ deriving instance Show AEvt
 
 type ConnInfo = ByteString
 
+type SndQueueSecured = Bool
+
 -- | Parameterized type for SMP agent events
 data AEvent (e :: AEntity) where
   INV :: AConnectionRequestUri -> AEvent AEConn
@@ -335,6 +344,7 @@ data AEvent (e :: AEntity) where
   INFO :: PQSupport -> ConnInfo -> AEvent AEConn
   CON :: PQEncryption -> AEvent AEConn -- notification that connection is established
   END :: AEvent AEConn
+  DELD :: AEvent AEConn
   CONNECT :: AProtocolType -> TransportHost -> AEvent AENone
   DISCONNECT :: AProtocolType -> TransportHost -> AEvent AENone
   DOWN :: SMPServer -> [ConnId] -> AEvent AENone
@@ -346,7 +356,7 @@ data AEvent (e :: AEntity) where
   MERR :: AgentMsgId -> AgentErrorType -> AEvent AEConn
   MERRS :: NonEmpty AgentMsgId -> AgentErrorType -> AEvent AEConn
   MSG :: MsgMeta -> MsgFlags -> MsgBody -> AEvent AEConn
-  MSGNTF :: SMPMsgMeta -> AEvent AEConn
+  MSGNTF :: MsgId -> Maybe UTCTime -> AEvent AEConn
   RCVD :: MsgMeta -> NonEmpty MsgReceipt -> AEvent AEConn
   QCONT :: AEvent AEConn
   DEL_RCVQ :: SMPServer -> SMP.RecipientId -> Maybe AgentErrorType -> AEvent AEConn
@@ -354,7 +364,9 @@ data AEvent (e :: AEntity) where
   DEL_USER :: Int64 -> AEvent AENone
   STAT :: ConnectionStats -> AEvent AEConn
   OK :: AEvent AEConn
+  JOINED :: SndQueueSecured -> AEvent AEConn
   ERR :: AgentErrorType -> AEvent AEConn
+  ERRS :: [(ConnId, AgentErrorType)] -> AEvent AENone
   SUSPENDED :: AEvent AENone
   RFPROG :: Int64 -> Int64 -> AEvent AERcvFile
   RFDONE :: FilePath -> AEvent AERcvFile
@@ -403,6 +415,7 @@ data AEventTag (e :: AEntity) where
   INFO_ :: AEventTag AEConn
   CON_ :: AEventTag AEConn
   END_ :: AEventTag AEConn
+  DELD_ :: AEventTag AEConn
   CONNECT_ :: AEventTag AENone
   DISCONNECT_ :: AEventTag AENone
   DOWN_ :: AEventTag AENone
@@ -422,7 +435,9 @@ data AEventTag (e :: AEntity) where
   DEL_USER_ :: AEventTag AENone
   STAT_ :: AEventTag AEConn
   OK_ :: AEventTag AEConn
+  JOINED_ :: AEventTag AEConn
   ERR_ :: AEventTag AEConn
+  ERRS_ :: AEventTag AENone
   SUSPENDED_ :: AEventTag AENone
   -- XFTP commands and responses
   RFDONE_ :: AEventTag AERcvFile
@@ -455,6 +470,7 @@ aEventTag = \case
   INFO {} -> INFO_
   CON _ -> CON_
   END -> END_
+  DELD -> DELD_
   CONNECT {} -> CONNECT_
   DISCONNECT {} -> DISCONNECT_
   DOWN {} -> DOWN_
@@ -474,7 +490,9 @@ aEventTag = \case
   DEL_USER _ -> DEL_USER_
   STAT _ -> STAT_
   OK -> OK_
+  JOINED _ -> JOINED_
   ERR _ -> ERR_
+  ERRS _ -> ERRS_
   SUSPENDED -> SUSPENDED_
   RFPROG {} -> RFPROG_
   RFDONE {} -> RFDONE_
@@ -1246,6 +1264,12 @@ instance Eq AConnectionRequestUri where
 
 deriving instance Show AConnectionRequestUri
 
+sameConnReqContact :: ConnectionRequestUri 'CMContact -> ConnectionRequestUri 'CMContact -> Bool
+sameConnReqContact (CRContactUri ConnReqUriData {crSmpQueues = qs}) (CRContactUri ConnReqUriData {crSmpQueues = qs'}) =
+  L.length qs == L.length qs' && all same (L.zip qs qs')
+  where
+    same (q, q') = sameQAddress (qAddress q) (qAddress q')
+
 data ConnReqUriData = ConnReqUriData
   { crScheme :: ServiceScheme,
     crAgentVRange :: VersionRangeSMPA,
@@ -1326,6 +1350,8 @@ data AgentErrorType
     CMD {cmdErr :: CommandErrorType, errContext :: String}
   | -- | connection errors
     CONN {connErr :: ConnectionErrorType}
+  | -- | user not found in database
+    NO_USER
   | -- | SMP protocol errors forwarded to agent clients
     SMP {serverAddress :: String, smpErr :: ErrorType}
   | -- | NTF protocol errors forwarded to agent clients

src/Simplex/Messaging/Agent/RetryInterval.hs

@@ -9,6 +9,7 @@ module Simplex.Messaging.Agent.RetryInterval
     RI2State (..),
     withRetryInterval,
     withRetryIntervalCount,
+    withRetryForeground,
     withRetryLock2,
     updateRetryInterval2,
     nextRetryDelay,
@@ -16,10 +17,11 @@ module Simplex.Messaging.Agent.RetryInterval
 where
 
 import Control.Concurrent (forkIO)
+import Control.Concurrent.STM (retry)
 import Control.Monad (void)
 import Control.Monad.IO.Class (MonadIO, liftIO)
 import Data.Int (Int64)
-import Simplex.Messaging.Util (threadDelay', whenM)
+import Simplex.Messaging.Util (threadDelay', unlessM, whenM)
 import UnliftIO.STM
 
 data RetryInterval = RetryInterval
@@ -63,6 +65,27 @@ withRetryIntervalCount ri action = callAction 0 0 $ initialInterval ri
           let elapsed' = elapsed + delay
           callAction (n + 1) elapsed' $ nextRetryDelay elapsed' delay ri
 
+withRetryForeground :: forall m a. MonadIO m => RetryInterval -> STM Bool -> STM Bool -> (Int64 -> m a -> m a) -> m a
+withRetryForeground ri isForeground isOnline action = callAction 0 $ initialInterval ri
+  where
+    callAction :: Int64 -> Int64 -> m a
+    callAction elapsed delay = action delay loop
+      where
+        loop = do
+          -- limit delay to max Int value (~36 minutes on for 32 bit architectures)
+          d <- registerDelay $ fromIntegral $ min delay (fromIntegral (maxBound :: Int))
+          (wasForeground, wasOnline) <- atomically $ (,) <$> isForeground <*> isOnline
+          reset <- atomically $ do
+            foreground <- isForeground
+            online <- isOnline
+            let reset = (not wasForeground && foreground) || (not wasOnline && online)
+            unlessM ((reset ||) <$> readTVar d) retry
+            pure reset
+          let (elapsed', delay')
+                | reset = (0, initialInterval ri)
+                | otherwise = (elapsed + delay, nextRetryDelay elapsed' delay ri)
+          callAction elapsed' delay'
+
 -- This function allows action to toggle between slow and fast retry intervals.
 withRetryLock2 :: forall m. MonadIO m => RetryInterval2 -> TMVar () -> (RI2State -> (RetryIntervalMode -> m ()) -> m ()) -> m ()
 withRetryLock2 RetryInterval2 {riSlow, riFast} lock action =

src/Simplex/Messaging/Agent/Stats.hs

@@ -1,16 +1,20 @@
 {-# LANGUAGE DuplicateRecordFields #-}
+{-# LANGUAGE GeneralizedNewtypeDeriving #-}
 {-# LANGUAGE NamedFieldPuns #-}
 {-# LANGUAGE TemplateHaskell #-}
 
 module Simplex.Messaging.Agent.Stats where
 
+import Data.Aeson (FromJSON (..), FromJSONKey, ToJSON (..))
 import qualified Data.Aeson.TH as J
-import Data.Map (Map)
+import Data.Int (Int64)
+import Data.Map.Strict (Map)
+import qualified Data.Map.Strict as M
 import Database.SQLite.Simple.FromField (FromField (..))
 import Database.SQLite.Simple.ToField (ToField (..))
 import Simplex.Messaging.Agent.Protocol (UserId)
 import Simplex.Messaging.Parsers (defaultJSON, fromTextField_)
-import Simplex.Messaging.Protocol (SMPServer, XFTPServer)
+import Simplex.Messaging.Protocol (SMPServer, XFTPServer, NtfServer)
 import Simplex.Messaging.Util (decodeJSON, encodeJSON)
 import UnliftIO.STM
 
@@ -29,13 +33,26 @@ data AgentSMPServerStats = AgentSMPServerStats
     recvDuplicates :: TVar Int, -- duplicate messages received
     recvCryptoErrs :: TVar Int, -- message decryption errors
     recvErrs :: TVar Int, -- receive errors
-    connCreated :: TVar Int,
-    connSecured :: TVar Int,
-    connCompleted :: TVar Int,
-    connDeleted :: TVar Int,
+    ackMsgs :: TVar Int, -- total messages acknowledged
+    ackAttempts :: TVar Int, -- acknowledgement attempts
+    ackNoMsgErrs :: TVar Int, -- NO_MSG ack errors
+    ackOtherErrs :: TVar Int, -- other permanent ack errors (temporary accounted for in attempts)
+    -- conn stats are accounted for rcv queue server
+    connCreated :: TVar Int, -- total connections created
+    connSecured :: TVar Int, -- connections secured
+    connCompleted :: TVar Int, -- connections completed
+    connDeleted :: TVar Int, -- total connections deleted
+    connDelAttempts :: TVar Int, -- total connection deletion attempts
+    connDelErrs :: TVar Int, -- permanent connection deletion errors (temporary accounted for in attempts)
     connSubscribed :: TVar Int, -- total successful subscription
     connSubAttempts :: TVar Int, -- subscription attempts
-    connSubErrs :: TVar Int -- permanent subscription errors (temporary accounted for in attempts)
+    connSubIgnored :: TVar Int, -- subscription results ignored (client switched to different session or it was not pending)
+    connSubErrs :: TVar Int, -- permanent subscription errors (temporary accounted for in attempts)
+    -- notifications stats
+    ntfKey :: TVar Int,
+    ntfKeyAttempts :: TVar Int,
+    ntfKeyDeleted :: TVar Int,
+    ntfKeyDeleteAttempts :: TVar Int
   }
 
 data AgentSMPServerStatsData = AgentSMPServerStatsData
@@ -53,16 +70,30 @@ data AgentSMPServerStatsData = AgentSMPServerStatsData
     _recvDuplicates :: Int,
     _recvCryptoErrs :: Int,
     _recvErrs :: Int,
+    _ackMsgs :: Int,
+    _ackAttempts :: Int,
+    _ackNoMsgErrs :: Int,
+    _ackOtherErrs :: Int,
     _connCreated :: Int,
     _connSecured :: Int,
     _connCompleted :: Int,
     _connDeleted :: Int,
+    _connDelAttempts :: Int,
+    _connDelErrs :: Int,
     _connSubscribed :: Int,
     _connSubAttempts :: Int,
-    _connSubErrs :: Int
+    _connSubIgnored :: Int,
+    _connSubErrs :: Int,
+    _ntfKey :: OptionalInt,
+    _ntfKeyAttempts :: OptionalInt,
+    _ntfKeyDeleted :: OptionalInt,
+    _ntfKeyDeleteAttempts :: OptionalInt
   }
   deriving (Show)
 
+newtype OptionalInt = OInt {toInt :: Int}
+  deriving (Num, Show, ToJSON)
+
 newAgentSMPServerStats :: STM AgentSMPServerStats
 newAgentSMPServerStats = do
   sentDirect <- newTVar 0
@@ -79,13 +110,24 @@ newAgentSMPServerStats = do
   recvDuplicates <- newTVar 0
   recvCryptoErrs <- newTVar 0
   recvErrs <- newTVar 0
+  ackMsgs <- newTVar 0
+  ackAttempts <- newTVar 0
+  ackNoMsgErrs <- newTVar 0
+  ackOtherErrs <- newTVar 0
   connCreated <- newTVar 0
   connSecured <- newTVar 0
   connCompleted <- newTVar 0
   connDeleted <- newTVar 0
+  connDelAttempts <- newTVar 0
+  connDelErrs <- newTVar 0
   connSubscribed <- newTVar 0
   connSubAttempts <- newTVar 0
+  connSubIgnored <- newTVar 0
   connSubErrs <- newTVar 0
+  ntfKey <- newTVar 0
+  ntfKeyAttempts <- newTVar 0
+  ntfKeyDeleted <- newTVar 0
+  ntfKeyDeleteAttempts <- newTVar 0
   pure
     AgentSMPServerStats
       { sentDirect,
@@ -102,15 +144,63 @@ newAgentSMPServerStats = do
         recvDuplicates,
         recvCryptoErrs,
         recvErrs,
+        ackMsgs,
+        ackAttempts,
+        ackNoMsgErrs,
+        ackOtherErrs,
         connCreated,
         connSecured,
         connCompleted,
         connDeleted,
+        connDelAttempts,
+        connDelErrs,
         connSubscribed,
         connSubAttempts,
-        connSubErrs
+        connSubIgnored,
+        connSubErrs,
+        ntfKey,
+        ntfKeyAttempts,
+        ntfKeyDeleted,
+        ntfKeyDeleteAttempts
       }
 
+newAgentSMPServerStatsData :: AgentSMPServerStatsData
+newAgentSMPServerStatsData =
+  AgentSMPServerStatsData
+    { _sentDirect = 0,
+      _sentViaProxy = 0,
+      _sentProxied = 0,
+      _sentDirectAttempts = 0,
+      _sentViaProxyAttempts = 0,
+      _sentProxiedAttempts = 0,
+      _sentAuthErrs = 0,
+      _sentQuotaErrs = 0,
+      _sentExpiredErrs = 0,
+      _sentOtherErrs = 0,
+      _recvMsgs = 0,
+      _recvDuplicates = 0,
+      _recvCryptoErrs = 0,
+      _recvErrs = 0,
+      _ackMsgs = 0,
+      _ackAttempts = 0,
+      _ackNoMsgErrs = 0,
+      _ackOtherErrs = 0,
+      _connCreated = 0,
+      _connSecured = 0,
+      _connCompleted = 0,
+      _connDeleted = 0,
+      _connDelAttempts = 0,
+      _connDelErrs = 0,
+      _connSubscribed = 0,
+      _connSubAttempts = 0,
+      _connSubIgnored = 0,
+      _connSubErrs = 0,
+      _ntfKey = 0,
+      _ntfKeyAttempts = 0,
+      _ntfKeyDeleted = 0,
+      _ntfKeyDeleteAttempts = 0
+    }
+
 newAgentSMPServerStats' :: AgentSMPServerStatsData -> STM AgentSMPServerStats
 newAgentSMPServerStats' s = do
   sentDirect <- newTVar $ _sentDirect s
@@ -127,13 +217,24 @@ newAgentSMPServerStats' s = do
   recvDuplicates <- newTVar $ _recvDuplicates s
   recvCryptoErrs <- newTVar $ _recvCryptoErrs s
   recvErrs <- newTVar $ _recvErrs s
+  ackMsgs <- newTVar $ _ackMsgs s
+  ackAttempts <- newTVar $ _ackAttempts s
+  ackNoMsgErrs <- newTVar $ _ackNoMsgErrs s
+  ackOtherErrs <- newTVar $ _ackOtherErrs s
   connCreated <- newTVar $ _connCreated s
   connSecured <- newTVar $ _connSecured s
   connCompleted <- newTVar $ _connCompleted s
   connDeleted <- newTVar $ _connDeleted s
+  connDelAttempts <- newTVar $ _connDelAttempts s
+  connDelErrs <- newTVar $ _connDelErrs s
   connSubscribed <- newTVar $ _connSubscribed s
   connSubAttempts <- newTVar $ _connSubAttempts s
+  connSubIgnored <- newTVar $ _connSubIgnored s
   connSubErrs <- newTVar $ _connSubErrs s
+  ntfKey <- newTVar $ toInt $ _ntfKey s
+  ntfKeyAttempts <- newTVar $ toInt $ _ntfKeyAttempts s
+  ntfKeyDeleted <- newTVar $ toInt $ _ntfKeyDeleted s
+  ntfKeyDeleteAttempts <- newTVar $ toInt $ _ntfKeyDeleteAttempts s
   pure
     AgentSMPServerStats
       { sentDirect,
@@ -150,13 +251,24 @@ newAgentSMPServerStats' s = do
         recvDuplicates,
         recvCryptoErrs,
         recvErrs,
+        ackMsgs,
+        ackAttempts,
+        ackNoMsgErrs,
+        ackOtherErrs,
         connCreated,
         connSecured,
         connCompleted,
         connDeleted,
+        connDelAttempts,
+        connDelErrs,
         connSubscribed,
         connSubAttempts,
-        connSubErrs
+        connSubIgnored,
+        connSubErrs,
+        ntfKey,
+        ntfKeyAttempts,
+        ntfKeyDeleted,
+        ntfKeyDeleteAttempts
       }
 
 -- as this is used to periodically update stats in db,
@@ -177,13 +289,24 @@ getAgentSMPServerStats s = do
   _recvDuplicates <- readTVarIO $ recvDuplicates s
   _recvCryptoErrs <- readTVarIO $ recvCryptoErrs s
   _recvErrs <- readTVarIO $ recvErrs s
+  _ackMsgs <- readTVarIO $ ackMsgs s
+  _ackAttempts <- readTVarIO $ ackAttempts s
+  _ackNoMsgErrs <- readTVarIO $ ackNoMsgErrs s
+  _ackOtherErrs <- readTVarIO $ ackOtherErrs s
   _connCreated <- readTVarIO $ connCreated s
   _connSecured <- readTVarIO $ connSecured s
   _connCompleted <- readTVarIO $ connCompleted s
   _connDeleted <- readTVarIO $ connDeleted s
+  _connDelAttempts <- readTVarIO $ connDelAttempts s
+  _connDelErrs <- readTVarIO $ connDelErrs s
   _connSubscribed <- readTVarIO $ connSubscribed s
   _connSubAttempts <- readTVarIO $ connSubAttempts s
+  _connSubIgnored <- readTVarIO $ connSubIgnored s
   _connSubErrs <- readTVarIO $ connSubErrs s
+  _ntfKey <- OInt <$> readTVarIO (ntfKey s)
+  _ntfKeyAttempts <- OInt <$> readTVarIO (ntfKeyAttempts s)
+  _ntfKeyDeleted <- OInt <$> readTVarIO (ntfKeyDeleted s)
+  _ntfKeyDeleteAttempts <- OInt <$> readTVarIO (ntfKeyDeleteAttempts s)
   pure
     AgentSMPServerStatsData
       { _sentDirect,
@@ -200,20 +323,70 @@ getAgentSMPServerStats s = do
         _recvDuplicates,
         _recvCryptoErrs,
         _recvErrs,
+        _ackMsgs,
+        _ackAttempts,
+        _ackNoMsgErrs,
+        _ackOtherErrs,
         _connCreated,
         _connSecured,
         _connCompleted,
         _connDeleted,
+        _connDelAttempts,
+        _connDelErrs,
         _connSubscribed,
         _connSubAttempts,
-        _connSubErrs
+        _connSubIgnored,
+        _connSubErrs,
+        _ntfKey,
+        _ntfKeyAttempts,
+        _ntfKeyDeleted,
+        _ntfKeyDeleteAttempts
       }
 
+addSMPStatsData :: AgentSMPServerStatsData -> AgentSMPServerStatsData -> AgentSMPServerStatsData
+addSMPStatsData sd1 sd2 =
+  AgentSMPServerStatsData
+    { _sentDirect = _sentDirect sd1 + _sentDirect sd2,
+      _sentViaProxy = _sentViaProxy sd1 + _sentViaProxy sd2,
+      _sentProxied = _sentProxied sd1 + _sentProxied sd2,
+      _sentDirectAttempts = _sentDirectAttempts sd1 + _sentDirectAttempts sd2,
+      _sentViaProxyAttempts = _sentViaProxyAttempts sd1 + _sentViaProxyAttempts sd2,
+      _sentProxiedAttempts = _sentProxiedAttempts sd1 + _sentProxiedAttempts sd2,
+      _sentAuthErrs = _sentAuthErrs sd1 + _sentAuthErrs sd2,
+      _sentQuotaErrs = _sentQuotaErrs sd1 + _sentQuotaErrs sd2,
+      _sentExpiredErrs = _sentExpiredErrs sd1 + _sentExpiredErrs sd2,
+      _sentOtherErrs = _sentOtherErrs sd1 + _sentOtherErrs sd2,
+      _recvMsgs = _recvMsgs sd1 + _recvMsgs sd2,
+      _recvDuplicates = _recvDuplicates sd1 + _recvDuplicates sd2,
+      _recvCryptoErrs = _recvCryptoErrs sd1 + _recvCryptoErrs sd2,
+      _recvErrs = _recvErrs sd1 + _recvErrs sd2,
+      _ackMsgs = _ackMsgs sd1 + _ackMsgs sd2,
+      _ackAttempts = _ackAttempts sd1 + _ackAttempts sd2,
+      _ackNoMsgErrs = _ackNoMsgErrs sd1 + _ackNoMsgErrs sd2,
+      _ackOtherErrs = _ackOtherErrs sd1 + _ackOtherErrs sd2,
+      _connCreated = _connCreated sd1 + _connCreated sd2,
+      _connSecured = _connSecured sd1 + _connSecured sd2,
+      _connCompleted = _connCompleted sd1 + _connCompleted sd2,
+      _connDeleted = _connDeleted sd1 + _connDeleted sd2,
+      _connDelAttempts = _connDelAttempts sd1 + _connDelAttempts sd2,
+      _connDelErrs = _connDelErrs sd1 + _connDelErrs sd2,
+      _connSubscribed = _connSubscribed sd1 + _connSubscribed sd2,
+      _connSubAttempts = _connSubAttempts sd1 + _connSubAttempts sd2,
+      _connSubIgnored = _connSubIgnored sd1 + _connSubIgnored sd2,
+      _connSubErrs = _connSubErrs sd1 + _connSubErrs sd2,
+      _ntfKey = _ntfKey sd1 + _ntfKey sd2,
+      _ntfKeyAttempts = _ntfKeyAttempts sd1 + _ntfKeyAttempts sd2,
+      _ntfKeyDeleted = _ntfKeyDeleted sd1 + _ntfKeyDeleted sd2,
+      _ntfKeyDeleteAttempts = _ntfKeyDeleteAttempts sd1 + _ntfKeyDeleteAttempts sd2
+    }
+
 data AgentXFTPServerStats = AgentXFTPServerStats
   { uploads :: TVar Int, -- total replicas uploaded to server
+    uploadsSize :: TVar Int64, -- total size of uploaded replicas in KB
     uploadAttempts :: TVar Int, -- upload attempts
     uploadErrs :: TVar Int, -- upload errors
     downloads :: TVar Int, -- total replicas downloaded from server
+    downloadsSize :: TVar Int64, -- total size of downloaded replicas in KB
     downloadAttempts :: TVar Int, -- download attempts
     downloadAuthErrs :: TVar Int, -- download AUTH errors
     downloadErrs :: TVar Int, -- other download errors (excluding above)
@@ -224,9 +397,11 @@ data AgentXFTPServerStats = AgentXFTPServerStats
 
 data AgentXFTPServerStatsData = AgentXFTPServerStatsData
   { _uploads :: Int,
+    _uploadsSize :: Int64,
     _uploadAttempts :: Int,
     _uploadErrs :: Int,
     _downloads :: Int,
+    _downloadsSize :: Int64,
     _downloadAttempts :: Int,
     _downloadAuthErrs :: Int,
     _downloadErrs :: Int,
@@ -239,9 +414,11 @@ data AgentXFTPServerStatsData = AgentXFTPServerStatsData
 newAgentXFTPServerStats :: STM AgentXFTPServerStats
 newAgentXFTPServerStats = do
   uploads <- newTVar 0
+  uploadsSize <- newTVar 0
   uploadAttempts <- newTVar 0
   uploadErrs <- newTVar 0
   downloads <- newTVar 0
+  downloadsSize <- newTVar 0
   downloadAttempts <- newTVar 0
   downloadAuthErrs <- newTVar 0
   downloadErrs <- newTVar 0
@@ -251,9 +428,11 @@ newAgentXFTPServerStats = do
   pure
     AgentXFTPServerStats
       { uploads,
+        uploadsSize,
         uploadAttempts,
         uploadErrs,
         downloads,
+        downloadsSize,
         downloadAttempts,
         downloadAuthErrs,
         downloadErrs,
@@ -262,12 +441,31 @@ newAgentXFTPServerStats = do
         deleteErrs
       }
 
+newAgentXFTPServerStatsData :: AgentXFTPServerStatsData
+newAgentXFTPServerStatsData =
+  AgentXFTPServerStatsData
+    { _uploads = 0,
+      _uploadsSize = 0,
+      _uploadAttempts = 0,
+      _uploadErrs = 0,
+      _downloads = 0,
+      _downloadsSize = 0,
+      _downloadAttempts = 0,
+      _downloadAuthErrs = 0,
+      _downloadErrs = 0,
+      _deletions = 0,
+      _deleteAttempts = 0,
+      _deleteErrs = 0
+    }
+
 newAgentXFTPServerStats' :: AgentXFTPServerStatsData -> STM AgentXFTPServerStats
 newAgentXFTPServerStats' s = do
   uploads <- newTVar $ _uploads s
+  uploadsSize <- newTVar $ _uploadsSize s
   uploadAttempts <- newTVar $ _uploadAttempts s
   uploadErrs <- newTVar $ _uploadErrs s
   downloads <- newTVar $ _downloads s
+  downloadsSize <- newTVar $ _downloadsSize s
   downloadAttempts <- newTVar $ _downloadAttempts s
   downloadAuthErrs <- newTVar $ _downloadAuthErrs s
   downloadErrs <- newTVar $ _downloadErrs s
@@ -277,9 +475,11 @@ newAgentXFTPServerStats' s = do
   pure
     AgentXFTPServerStats
       { uploads,
+        uploadsSize,
         uploadAttempts,
         uploadErrs,
         downloads,
+        downloadsSize,
         downloadAttempts,
         downloadAuthErrs,
         downloadErrs,
@@ -293,9 +493,11 @@ newAgentXFTPServerStats' s = do
 getAgentXFTPServerStats :: AgentXFTPServerStats -> IO AgentXFTPServerStatsData
 getAgentXFTPServerStats s = do
   _uploads <- readTVarIO $ uploads s
+  _uploadsSize <- readTVarIO $ uploadsSize s
   _uploadAttempts <- readTVarIO $ uploadAttempts s
   _uploadErrs <- readTVarIO $ uploadErrs s
   _downloads <- readTVarIO $ downloads s
+  _downloadsSize <- readTVarIO $ downloadsSize s
   _downloadAttempts <- readTVarIO $ downloadAttempts s
   _downloadAuthErrs <- readTVarIO $ downloadAuthErrs s
   _downloadErrs <- readTVarIO $ downloadErrs s
@@ -305,9 +507,11 @@ getAgentXFTPServerStats s = do
   pure
     AgentXFTPServerStatsData
       { _uploads,
+        _uploadsSize,
         _uploadAttempts,
         _uploadErrs,
         _downloads,
+        _downloadsSize,
         _downloadAttempts,
         _downloadAuthErrs,
         _downloadErrs,
@@ -316,18 +520,144 @@ getAgentXFTPServerStats s = do
         _deleteErrs
       }
 
+addXFTPStatsData :: AgentXFTPServerStatsData -> AgentXFTPServerStatsData -> AgentXFTPServerStatsData
+addXFTPStatsData sd1 sd2 =
+  AgentXFTPServerStatsData
+    { _uploads = _uploads sd1 + _uploads sd2,
+      _uploadsSize = _uploadsSize sd1 + _uploadsSize sd2,
+      _uploadAttempts = _uploadAttempts sd1 + _uploadAttempts sd2,
+      _uploadErrs = _uploadErrs sd1 + _uploadErrs sd2,
+      _downloads = _downloads sd1 + _downloads sd2,
+      _downloadsSize = _downloadsSize sd1 + _downloadsSize sd2,
+      _downloadAttempts = _downloadAttempts sd1 + _downloadAttempts sd2,
+      _downloadAuthErrs = _downloadAuthErrs sd1 + _downloadAuthErrs sd2,
+      _downloadErrs = _downloadErrs sd1 + _downloadErrs sd2,
+      _deletions = _deletions sd1 + _deletions sd2,
+      _deleteAttempts = _deleteAttempts sd1 + _deleteAttempts sd2,
+      _deleteErrs = _deleteErrs sd1 + _deleteErrs sd2
+    }
+
+data AgentNtfServerStats = AgentNtfServerStats
+  { ntfCreated :: TVar Int,
+    ntfCreateAttempts :: TVar Int,
+    ntfChecked :: TVar Int,
+    ntfCheckAttempts :: TVar Int,
+    ntfDeleted :: TVar Int,
+    ntfDelAttempts :: TVar Int
+  }
+
+data AgentNtfServerStatsData = AgentNtfServerStatsData
+  { _ntfCreated :: Int,
+    _ntfCreateAttempts :: Int,
+    _ntfChecked :: Int,
+    _ntfCheckAttempts :: Int,
+    _ntfDeleted :: Int,
+    _ntfDelAttempts :: Int
+  }
+  deriving (Show)
+
+newAgentNtfServerStats :: STM AgentNtfServerStats
+newAgentNtfServerStats = do
+  ntfCreated <- newTVar 0
+  ntfCreateAttempts <- newTVar 0
+  ntfChecked <- newTVar 0
+  ntfCheckAttempts <- newTVar 0
+  ntfDeleted <- newTVar 0
+  ntfDelAttempts <- newTVar 0
+  pure
+    AgentNtfServerStats
+      { ntfCreated,
+        ntfCreateAttempts,
+        ntfChecked,
+        ntfCheckAttempts,
+        ntfDeleted,
+        ntfDelAttempts
+      }
+
+newAgentNtfServerStatsData :: AgentNtfServerStatsData
+newAgentNtfServerStatsData =
+  AgentNtfServerStatsData
+    { _ntfCreated = 0,
+      _ntfCreateAttempts = 0,
+      _ntfChecked = 0,
+      _ntfCheckAttempts = 0,
+      _ntfDeleted = 0,
+      _ntfDelAttempts = 0
+    }
+
+newAgentNtfServerStats' :: AgentNtfServerStatsData -> STM AgentNtfServerStats
+newAgentNtfServerStats' s = do
+  ntfCreated <- newTVar $ _ntfCreated s
+  ntfCreateAttempts <- newTVar $ _ntfCreateAttempts s
+  ntfChecked <- newTVar $ _ntfChecked s
+  ntfCheckAttempts <- newTVar $ _ntfCheckAttempts s
+  ntfDeleted <- newTVar $ _ntfDeleted s
+  ntfDelAttempts <- newTVar $ _ntfDelAttempts s
+  pure
+    AgentNtfServerStats
+      { ntfCreated,
+        ntfCreateAttempts,
+        ntfChecked,
+        ntfCheckAttempts,
+        ntfDeleted,
+        ntfDelAttempts
+      }
+
+getAgentNtfServerStats :: AgentNtfServerStats -> IO AgentNtfServerStatsData
+getAgentNtfServerStats s = do
+  _ntfCreated <- readTVarIO $ ntfCreated s
+  _ntfCreateAttempts <- readTVarIO $ ntfCreateAttempts s
+  _ntfChecked <- readTVarIO $ ntfChecked s
+  _ntfCheckAttempts <- readTVarIO $ ntfCheckAttempts s
+  _ntfDeleted <- readTVarIO $ ntfDeleted s
+  _ntfDelAttempts <- readTVarIO $ ntfDelAttempts s
+  pure
+    AgentNtfServerStatsData
+      { _ntfCreated,
+        _ntfCreateAttempts,
+        _ntfChecked,
+        _ntfCheckAttempts,
+        _ntfDeleted,
+        _ntfDelAttempts
+      }
+
+addNtfStatsData :: AgentNtfServerStatsData -> AgentNtfServerStatsData -> AgentNtfServerStatsData
+addNtfStatsData sd1 sd2 =
+  AgentNtfServerStatsData
+    { _ntfCreated = _ntfCreated sd1 + _ntfCreated sd2,
+      _ntfCreateAttempts = _ntfCreateAttempts sd1 + _ntfCreateAttempts sd2,
+      _ntfChecked = _ntfChecked sd1 + _ntfChecked sd2,
+      _ntfCheckAttempts = _ntfCheckAttempts sd1 + _ntfCheckAttempts sd2,
+      _ntfDeleted = _ntfDeleted sd1 + _ntfDeleted sd2,
+      _ntfDelAttempts = _ntfDelAttempts sd1 + _ntfDelAttempts sd2
+    }
+
 -- Type for gathering both smp and xftp stats across all users and servers,
 -- to then be persisted to db as a single json.
 data AgentPersistedServerStats = AgentPersistedServerStats
   { smpServersStats :: Map (UserId, SMPServer) AgentSMPServerStatsData,
-    xftpServersStats :: Map (UserId, XFTPServer) AgentXFTPServerStatsData
+    xftpServersStats :: Map (UserId, XFTPServer) AgentXFTPServerStatsData,
+    ntfServersStats :: OptionalMap (UserId, NtfServer) AgentNtfServerStatsData
   }
   deriving (Show)
 
+instance FromJSON OptionalInt where
+  parseJSON v = OInt <$> parseJSON v
+  omittedField = Just (OInt 0)
+
+newtype OptionalMap k v = OptionalMap (Map k v)
+  deriving (Show, ToJSON)
+
+instance (FromJSONKey k, Ord k, FromJSON v) => FromJSON (OptionalMap k v) where
+  parseJSON v = OptionalMap <$> parseJSON v
+  omittedField = Just (OptionalMap M.empty)
+
 $(J.deriveJSON defaultJSON ''AgentSMPServerStatsData)
 
 $(J.deriveJSON defaultJSON ''AgentXFTPServerStatsData)
 
+$(J.deriveJSON defaultJSON ''AgentNtfServerStatsData)
+
 $(J.deriveJSON defaultJSON ''AgentPersistedServerStats)
 
 instance ToField AgentPersistedServerStats where

src/Simplex/Messaging/Agent/Store.hs

@@ -175,6 +175,12 @@ instance SMPQueue RcvQueue where
   queueId RcvQueue {rcvId} = rcvId
   {-# INLINE queueId #-}
 
+instance SMPQueue NewRcvQueue where
+  qServer RcvQueue {server} = server
+  {-# INLINE qServer #-}
+  queueId RcvQueue {rcvId} = rcvId
+  {-# INLINE queueId #-}
+
 instance SMPQueue SndQueue where
   qServer SndQueue {server} = server
   {-# INLINE qServer #-}
@@ -617,7 +623,7 @@ data StoreError
   | -- | Confirmation not found.
     SEConfirmationNotFound
   | -- | Invitation not found
-    SEInvitationNotFound
+    SEInvitationNotFound String InvitationId
   | -- | Message not found
     SEMsgNotFound
   | -- | Command not found

src/Simplex/Messaging/Agent/Store/SQLite.hs

@@ -1,10 +1,12 @@
 {-# LANGUAGE ConstraintKinds #-}
 {-# LANGUAGE DataKinds #-}
 {-# LANGUAGE DeriveAnyClass #-}
+{-# LANGUAGE DerivingStrategies #-}
 {-# LANGUAGE DuplicateRecordFields #-}
 {-# LANGUAGE FlexibleContexts #-}
 {-# LANGUAGE FlexibleInstances #-}
 {-# LANGUAGE GADTs #-}
+{-# LANGUAGE GeneralizedNewtypeDeriving #-}
 {-# LANGUAGE InstanceSigs #-}
 {-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE MultiParamTypeClasses #-}
@@ -15,6 +17,7 @@
 {-# LANGUAGE QuasiQuotes #-}
 {-# LANGUAGE RecordWildCards #-}
 {-# LANGUAGE ScopedTypeVariables #-}
+{-# LANGUAGE StandaloneDeriving #-}
 {-# LANGUAGE TemplateHaskell #-}
 {-# LANGUAGE TupleSections #-}
 {-# LANGUAGE TypeOperators #-}
@@ -57,6 +60,7 @@ module Simplex.Messaging.Agent.Store.SQLite
     getDeletedConns,
     getConnData,
     setConnDeleted,
+    setConnUserId,
     setConnAgentVersion,
     setConnPQSupport,
     getDeletedConnIds,
@@ -110,6 +114,7 @@ module Simplex.Messaging.Agent.Store.SQLite
     getSndMsgViaRcpt,
     updateSndMsgRcpt,
     getPendingQueueMsg,
+    getConnectionsForDelivery,
     updatePendingMsgRIState,
     deletePendingMsgs,
     getExpiredSndMessages,
@@ -117,6 +122,7 @@ module Simplex.Messaging.Agent.Store.SQLite
     getRcvMsg,
     getLastMsg,
     checkRcvMsgHashExists,
+    getRcvMsgBrokerTs,
     deleteMsg,
     deleteDeliveredSndMsg,
     deleteSndMsgDelivery,
@@ -135,6 +141,7 @@ module Simplex.Messaging.Agent.Store.SQLite
     createCommand,
     getPendingCommandServers,
     getPendingServerCommand,
+    updateCommandServer,
     deleteCommand,
     -- Notification device token persistence
     createNtfToken,
@@ -144,7 +151,15 @@ module Simplex.Messaging.Agent.Store.SQLite
     updateNtfMode,
     updateNtfToken,
     removeNtfToken,
+    addNtfTokenToDelete,
+    deleteExpiredNtfTokensToDelete,
+    NtfTokenToDelete,
+    getNextNtfTokenToDelete,
+    markNtfTokenToDeleteFailed_, -- exported for tests
+    getPendingDelTknServers,
+    deleteNtfTokenToDelete,
     -- Notification subscription persistence
+    NtfSupervisorSub,
     getNtfSubscription,
     createNtfSubscription,
     supervisorUpdateNtfSub,
@@ -152,9 +167,10 @@ module Simplex.Messaging.Agent.Store.SQLite
     updateNtfSubscription,
     setNullNtfSubscriptionAction,
     deleteNtfSubscription,
-    getNextNtfSubNTFAction,
+    deleteNtfSubscription',
+    getNextNtfSubNTFActions,
     markNtfSubActionNtfFailed_, -- exported for tests
-    getNextNtfSubSMPAction,
+    getNextNtfSubSMPActions,
     markNtfSubActionSMPFailed_, -- exported for tests
     getActiveNtfToken,
     getNtfRcvQueue,
@@ -220,7 +236,7 @@ module Simplex.Messaging.Agent.Store.SQLite
     -- * utilities
     withConnection,
     withTransaction,
-    withTransactionCtx,
+    withTransactionPriority,
     firstRow,
     firstRow',
     maybeFirstRow,
@@ -392,10 +408,10 @@ connectSQLiteStore dbFilePath key keepKey = do
   dbNew <- not <$> doesFileExist dbFilePath
   dbConn <- dbBusyLoop (connectDB dbFilePath key)
   dbConnection <- newMVar dbConn
-  atomically $ do
-    dbKey <- newTVar $! storeKey key keepKey
-    dbClosed <- newTVar False
-    pure SQLiteStore {dbFilePath, dbKey, dbConnection, dbNew, dbClosed}
+  dbKey <- newTVarIO $! storeKey key keepKey
+  dbClosed <- newTVarIO False
+  dbSem <- newTVarIO 0
+  pure SQLiteStore {dbFilePath, dbKey, dbSem, dbConnection, dbNew, dbClosed}
 
 connectDB :: FilePath -> ScrubbedBytes -> IO DB.Connection
 connectDB path key = do
@@ -910,9 +926,9 @@ createInvitation db gVar NewInvitation {contactConnId, connReq, recipientConnInf
       |]
       (invitationId, contactConnId, connReq, recipientConnInfo)
 
-getInvitation :: DB.Connection -> InvitationId -> IO (Either StoreError Invitation)
-getInvitation db invitationId =
-  firstRow invitation SEInvitationNotFound $
+getInvitation :: DB.Connection -> String -> InvitationId -> IO (Either StoreError Invitation)
+getInvitation db cxt invitationId =
+  firstRow invitation (SEInvitationNotFound cxt invitationId) $
     DB.query
       db
       [sql|
@@ -960,17 +976,18 @@ updateRcvIds db connId = do
   pure (internalId, internalRcvId, lastExternalSndId, lastRcvHash)
 
 createRcvMsg :: DB.Connection -> ConnId -> RcvQueue -> RcvMsgData -> IO ()
-createRcvMsg db connId rq rcvMsgData@RcvMsgData {msgMeta = MsgMeta {sndMsgId}, internalRcvId, internalHash} = do
+createRcvMsg db connId rq@RcvQueue {dbQueueId} rcvMsgData@RcvMsgData {msgMeta = MsgMeta {sndMsgId, broker = (_, brokerTs)}, internalRcvId, internalHash} = do
   insertRcvMsgBase_ db connId rcvMsgData
   insertRcvMsgDetails_ db connId rq rcvMsgData
   updateRcvMsgHash db connId sndMsgId internalRcvId internalHash
+  DB.execute db "UPDATE rcv_queues SET last_broker_ts = ? WHERE conn_id = ? AND rcv_queue_id = ?" (brokerTs, connId, dbQueueId)
 
-updateSndIds :: DB.Connection -> ConnId -> IO (InternalId, InternalSndId, PrevSndMsgHash)
-updateSndIds db connId = do
-  (lastInternalId, lastInternalSndId, prevSndHash) <- retrieveLastIdsAndHashSnd_ db connId
+updateSndIds :: DB.Connection -> ConnId -> IO (Either StoreError (InternalId, InternalSndId, PrevSndMsgHash))
+updateSndIds db connId = runExceptT $ do
+  (lastInternalId, lastInternalSndId, prevSndHash) <- ExceptT $ retrieveLastIdsAndHashSnd_ db connId
   let internalId = InternalId $ unId lastInternalId + 1
       internalSndId = InternalSndId $ unSndId lastInternalSndId + 1
-  updateLastIdsSnd_ db connId internalId internalSndId
+  liftIO $ updateLastIdsSnd_ db connId internalId internalSndId
   pure (internalId, internalSndId, prevSndHash)
 
 createSndMsg :: DB.Connection -> ConnId -> SndMsgData -> IO ()
@@ -1008,6 +1025,10 @@ updateSndMsgRcpt db connId sndMsgId MsgReceipt {agentMsgId, msgRcptStatus} =
     "UPDATE snd_messages SET rcpt_internal_id = ?, rcpt_status = ? WHERE conn_id = ? AND internal_snd_id = ?"
     (agentMsgId, msgRcptStatus, connId, sndMsgId)
 
+getConnectionsForDelivery :: DB.Connection -> IO [ConnId]
+getConnectionsForDelivery db =
+  map fromOnly <$> DB.query_ db "SELECT DISTINCT conn_id FROM snd_message_deliveries WHERE failed = 0"
+
 getPendingQueueMsg :: DB.Connection -> ConnId -> SndQueue -> IO (Either StoreError (Maybe (Maybe RcvQueue, PendingMsgData)))
 getPendingQueueMsg db connId SndQueue {dbQueueId} =
   getWorkItem "message" getMsgId getMsgData markMsgFailed
@@ -1051,17 +1072,26 @@ getPendingQueueMsg db connId SndQueue {dbQueueId} =
 
 getWorkItem :: Show i => ByteString -> IO (Maybe i) -> (i -> IO (Either StoreError a)) -> (i -> IO ()) -> IO (Either StoreError (Maybe a))
 getWorkItem itemName getId getItem markFailed =
-  runExceptT $ handleErr "getId" getId >>= mapM tryGetItem
+  runExceptT $ handleWrkErr itemName "getId" getId >>= mapM (tryGetItem itemName getItem markFailed)
+
+getWorkItems :: Show i => ByteString -> IO [i] -> (i -> IO (Either StoreError a)) -> (i -> IO ()) -> IO (Either StoreError [Either StoreError a])
+getWorkItems itemName getIds getItem markFailed =
+  runExceptT $ handleWrkErr itemName "getIds" getIds >>= mapM (tryE . tryGetItem itemName getItem markFailed)
+
+tryGetItem :: Show i => ByteString -> (i -> IO (Either StoreError a)) -> (i -> IO ()) -> i -> ExceptT StoreError IO a
+tryGetItem itemName getItem markFailed itemId = ExceptT (getItem itemId) `catchStoreError` \e -> mark >> throwE e
   where
-    tryGetItem itemId = ExceptT (getItem itemId) `catchStoreErrors` \e -> mark itemId >> throwE e
-    mark itemId = handleErr ("markFailed ID " <> bshow itemId) $ markFailed itemId
-    catchStoreErrors = catchAllErrors (SEInternal . bshow)
-    -- Errors caught by this function will suspend worker as if there is no more work,
-    handleErr :: ByteString -> IO a -> ExceptT StoreError IO a
-    handleErr opName action = ExceptT $ first mkError <$> E.try action
-      where
-        mkError :: E.SomeException -> StoreError
-        mkError e = SEWorkItemError $ itemName <> " " <> opName <> " error: " <> bshow e
+    mark = handleWrkErr itemName ("markFailed ID " <> bshow itemId) $ markFailed itemId
+
+catchStoreError :: ExceptT StoreError IO a -> (StoreError -> ExceptT StoreError IO a) -> ExceptT StoreError IO a
+catchStoreError = catchAllErrors (SEInternal . bshow)
+
+-- Errors caught by this function will suspend worker as if there is no more work,
+handleWrkErr :: ByteString -> ByteString -> IO a -> ExceptT StoreError IO a
+handleWrkErr itemName opName action = ExceptT $ first mkError <$> E.try action
+  where
+    mkError :: E.SomeException -> StoreError
+    mkError e = SEWorkItemError $ itemName <> " " <> opName <> " error: " <> bshow e
 
 updatePendingMsgRIState :: DB.Connection -> ConnId -> InternalId -> RI2State -> IO ()
 updatePendingMsgRIState db connId msgId RI2State {slowInterval, fastInterval} =
@@ -1157,6 +1187,11 @@ checkRcvMsgHashExists db connId hash = do
           (connId, hash)
       )
 
+getRcvMsgBrokerTs :: DB.Connection -> ConnId -> SMP.MsgId -> IO (Either StoreError BrokerTs)
+getRcvMsgBrokerTs db connId msgId =
+  firstRow fromOnly SEMsgNotFound $
+    DB.query db "SELECT broker_ts FROM rcv_messages WHERE conn_id = ? AND broker_id = ?" (connId, msgId)
+
 deleteMsg :: DB.Connection -> ConnId -> InternalId -> IO ()
 deleteMsg db connId msgId =
   DB.execute db "DELETE FROM messages WHERE conn_id = ? AND internal_id = ?;" (connId, msgId)
@@ -1317,38 +1352,39 @@ getPendingCommandServers db connId = do
   where
     smpServer (host, port, keyHash) = SMPServer <$> host <*> port <*> keyHash
 
-getPendingServerCommand :: DB.Connection -> Maybe SMPServer -> IO (Either StoreError (Maybe PendingCommand))
-getPendingServerCommand db srv_ = getWorkItem "command" getCmdId getCommand markCommandFailed
+getPendingServerCommand :: DB.Connection -> ConnId -> Maybe SMPServer -> IO (Either StoreError (Maybe PendingCommand))
+getPendingServerCommand db connId srv_ = getWorkItem "command" getCmdId getCommand markCommandFailed
   where
     getCmdId :: IO (Maybe Int64)
     getCmdId =
       maybeFirstRow fromOnly $ case srv_ of
         Nothing ->
-          DB.query_
+          DB.query
             db
             [sql|
               SELECT command_id FROM commands
-              WHERE host IS NULL AND port IS NULL AND failed = 0
+              WHERE conn_id = ? AND host IS NULL AND port IS NULL AND failed = 0
               ORDER BY created_at ASC, command_id ASC
               LIMIT 1
             |]
+            (Only connId)
         Just (SMPServer host port _) ->
           DB.query
             db
             [sql|
               SELECT command_id FROM commands
-              WHERE host = ? AND port = ? AND failed = 0
+              WHERE conn_id = ? AND host = ? AND port = ? AND failed = 0
               ORDER BY created_at ASC, command_id ASC
               LIMIT 1
             |]
-            (host, port)
+            (connId, host, port)
     getCommand :: Int64 -> IO (Either StoreError PendingCommand)
     getCommand cmdId =
       firstRow pendingCommand err $
         DB.query
           db
           [sql|
-            SELECT c.corr_id, cs.user_id, c.conn_id, c.command
+            SELECT c.corr_id, cs.user_id, c.command
             FROM commands c
             JOIN connections cs USING (conn_id)
             WHERE c.command_id = ?
@@ -1356,9 +1392,22 @@ getPendingServerCommand db srv_ = getWorkItem "command" getCmdId getCommand mark
           (Only cmdId)
       where
         err = SEInternal $ "command  " <> bshow cmdId <> " returned []"
-        pendingCommand (corrId, userId, connId, command) = PendingCommand {cmdId, corrId, userId, connId, command}
+        pendingCommand (corrId, userId, command) = PendingCommand {cmdId, corrId, userId, connId, command}
     markCommandFailed cmdId = DB.execute db "UPDATE commands SET failed = 1 WHERE command_id = ?" (Only cmdId)
 
+updateCommandServer :: DB.Connection -> AsyncCmdId -> SMPServer -> IO (Either StoreError ())
+updateCommandServer db cmdId srv@(SMPServer host port _) = runExceptT $ do
+  serverKeyHash_ <- ExceptT $ getServerKeyHash_ db srv
+  liftIO $
+    DB.execute
+      db
+      [sql|
+        UPDATE commands
+        SET host = ?, port = ?, server_key_hash = ?
+        WHERE command_id = ?
+      |]
+      (host, port, serverKeyHash_, cmdId)
+
 deleteCommand :: DB.Connection -> AsyncCmdId -> IO ()
 deleteCommand db cmdId =
   DB.execute db "DELETE FROM commands WHERE command_id = ?" (Only cmdId)
@@ -1451,29 +1500,96 @@ removeNtfToken db NtfToken {deviceToken = DeviceToken provider token, ntfServer
     |]
     (provider, token, host, port)
 
-getNtfSubscription :: DB.Connection -> ConnId -> IO (Maybe (NtfSubscription, Maybe (NtfSubAction, NtfActionTs)))
+addNtfTokenToDelete :: DB.Connection -> NtfServer -> C.APrivateAuthKey -> NtfTokenId -> IO ()
+addNtfTokenToDelete db ProtocolServer {host, port, keyHash} ntfPrivKey tknId =
+  DB.execute db "INSERT INTO ntf_tokens_to_delete (ntf_host, ntf_port, ntf_key_hash, tkn_id, tkn_priv_key) VALUES (?,?,?,?,?)" (host, port, keyHash, tknId, ntfPrivKey)
+
+deleteExpiredNtfTokensToDelete :: DB.Connection -> NominalDiffTime -> IO ()
+deleteExpiredNtfTokensToDelete db ttl = do
+  cutoffTs <- addUTCTime (-ttl) <$> getCurrentTime
+  DB.execute db "DELETE FROM ntf_tokens_to_delete WHERE created_at < ?" (Only cutoffTs)
+
+type NtfTokenToDelete = (Int64, C.APrivateAuthKey, NtfTokenId)
+
+getNextNtfTokenToDelete :: DB.Connection -> NtfServer -> IO (Either StoreError (Maybe NtfTokenToDelete))
+getNextNtfTokenToDelete db (NtfServer ntfHost ntfPort _) =
+  getWorkItem "ntf tkn del" getNtfTknDbId getNtfTknToDelete (markNtfTokenToDeleteFailed_ db)
+  where
+    getNtfTknDbId :: IO (Maybe Int64)
+    getNtfTknDbId =
+      maybeFirstRow fromOnly $
+        DB.query
+          db
+          [sql|
+            SELECT ntf_token_to_delete_id
+            FROM ntf_tokens_to_delete
+            WHERE ntf_host = ? AND ntf_port = ?
+              AND del_failed = 0
+            ORDER BY created_at ASC
+            LIMIT 1
+          |]
+          (ntfHost, ntfPort)
+    getNtfTknToDelete :: Int64 -> IO (Either StoreError NtfTokenToDelete)
+    getNtfTknToDelete tknDbId =
+      firstRow ntfTokenToDelete err $
+        DB.query
+          db
+          [sql|
+            SELECT tkn_priv_key, tkn_id
+            FROM ntf_tokens_to_delete
+            WHERE ntf_token_to_delete_id = ?
+          |]
+          (Only tknDbId)
+      where
+        err = SEInternal $ "ntf token to delete " <> bshow tknDbId <> " returned []"
+        ntfTokenToDelete (tknPrivKey, tknId) = (tknDbId, tknPrivKey, tknId)
+
+markNtfTokenToDeleteFailed_ :: DB.Connection -> Int64 -> IO ()
+markNtfTokenToDeleteFailed_ db tknDbId =
+  DB.execute db "UPDATE ntf_tokens_to_delete SET del_failed = 1 where ntf_token_to_delete_id = ?" (Only tknDbId)
+
+getPendingDelTknServers :: DB.Connection -> IO [NtfServer]
+getPendingDelTknServers db =
+  map toNtfServer
+    <$> DB.query_
+      db
+      [sql|
+        SELECT DISTINCT ntf_host, ntf_port, ntf_key_hash
+        FROM ntf_tokens_to_delete
+      |]
+  where
+    toNtfServer (host, port, keyHash) = NtfServer host port keyHash
+
+deleteNtfTokenToDelete :: DB.Connection -> Int64 -> IO ()
+deleteNtfTokenToDelete db tknDbId =
+  DB.execute db "DELETE FROM ntf_tokens_to_delete WHERE ntf_token_to_delete_id = ?" (Only tknDbId)
+
+type NtfSupervisorSub = (NtfSubscription, Maybe (NtfSubAction, NtfActionTs))
+
+getNtfSubscription :: DB.Connection -> ConnId -> IO (Maybe NtfSupervisorSub)
 getNtfSubscription db connId =
   maybeFirstRow ntfSubscription $
     DB.query
       db
       [sql|
-        SELECT s.host, s.port, COALESCE(nsb.smp_server_key_hash, s.key_hash), ns.ntf_host, ns.ntf_port, ns.ntf_key_hash,
+        SELECT c.user_id, s.host, s.port, COALESCE(nsb.smp_server_key_hash, s.key_hash), ns.ntf_host, ns.ntf_port, ns.ntf_key_hash,
           nsb.smp_ntf_id, nsb.ntf_sub_id, nsb.ntf_sub_status, nsb.ntf_sub_action, nsb.ntf_sub_smp_action, nsb.ntf_sub_action_ts
         FROM ntf_subscriptions nsb
+        JOIN connections c USING (conn_id)
         JOIN servers s ON s.host = nsb.smp_host AND s.port = nsb.smp_port
         JOIN ntf_servers ns USING (ntf_host, ntf_port)
         WHERE nsb.conn_id = ?
       |]
       (Only connId)
   where
-    ntfSubscription (smpHost, smpPort, smpKeyHash, ntfHost, ntfPort, ntfKeyHash, ntfQueueId, ntfSubId, ntfSubStatus, ntfAction_, smpAction_, actionTs_) =
+    ntfSubscription ((userId, smpHost, smpPort, smpKeyHash, ntfHost, ntfPort, ntfKeyHash) :. (ntfQueueId, ntfSubId, ntfSubStatus, ntfAction_, smpAction_, actionTs_)) =
       let smpServer = SMPServer smpHost smpPort smpKeyHash
           ntfServer = NtfServer ntfHost ntfPort ntfKeyHash
           action = case (ntfAction_, smpAction_, actionTs_) of
-            (Just ntfAction, Nothing, Just actionTs) -> Just (NtfSubNTFAction ntfAction, actionTs)
-            (Nothing, Just smpAction, Just actionTs) -> Just (NtfSubSMPAction smpAction, actionTs)
+            (Just ntfAction, Nothing, Just actionTs) -> Just (NSANtf ntfAction, actionTs)
+            (Nothing, Just smpAction, Just actionTs) -> Just (NSASMP smpAction, actionTs)
             _ -> Nothing
-       in (NtfSubscription {connId, smpServer, ntfQueueId, ntfServer, ntfSubId, ntfSubStatus}, action)
+       in (NtfSubscription {userId, connId, smpServer, ntfQueueId, ntfServer, ntfSubId, ntfSubStatus}, action)
 
 createNtfSubscription :: DB.Connection -> NtfSubscription -> NtfSubAction -> IO (Either StoreError ())
 createNtfSubscription db ntfSubscription action = runExceptT $ do
@@ -1496,16 +1612,19 @@ createNtfSubscription db ntfSubscription action = runExceptT $ do
     (ntfSubAction, ntfSubSMPAction) = ntfSubAndSMPAction action
 
 supervisorUpdateNtfSub :: DB.Connection -> NtfSubscription -> NtfSubAction -> IO ()
-supervisorUpdateNtfSub db NtfSubscription {connId, ntfQueueId, ntfServer = (NtfServer ntfHost ntfPort _), ntfSubId, ntfSubStatus} action = do
+supervisorUpdateNtfSub db NtfSubscription {connId, smpServer = (SMPServer smpHost smpPort _), ntfQueueId, ntfServer = (NtfServer ntfHost ntfPort _), ntfSubId, ntfSubStatus} action = do
   ts <- getCurrentTime
   DB.execute
     db
     [sql|
       UPDATE ntf_subscriptions
-      SET smp_ntf_id = ?, ntf_host = ?, ntf_port = ?, ntf_sub_id = ?, ntf_sub_status = ?, ntf_sub_action = ?, ntf_sub_smp_action = ?, ntf_sub_action_ts = ?, updated_by_supervisor = ?, updated_at = ?
+      SET smp_host = ?, smp_port = ?, smp_ntf_id = ?, ntf_host = ?, ntf_port = ?, ntf_sub_id = ?,
+          ntf_sub_status = ?, ntf_sub_action = ?, ntf_sub_smp_action = ?, ntf_sub_action_ts = ?, updated_by_supervisor = ?, updated_at = ?
       WHERE conn_id = ?
     |]
-    ((ntfQueueId, ntfHost, ntfPort, ntfSubId) :. (ntfSubStatus, ntfSubAction, ntfSubSMPAction, ts, True, ts, connId))
+    ( (smpHost, smpPort, ntfQueueId, ntfHost, ntfPort, ntfSubId)
+        :. (ntfSubStatus, ntfSubAction, ntfSubSMPAction, ts, True, ts, connId)
+    )
   where
     (ntfSubAction, ntfSubSMPAction) = ntfSubAndSMPAction action
 
@@ -1580,16 +1699,20 @@ deleteNtfSubscription db connId = do
             WHERE conn_id = ?
           |]
           (Nothing :: Maybe SMP.NotifierId, Nothing :: Maybe NtfSubscriptionId, NASDeleted, False, updatedAt, connId)
-      else DB.execute db "DELETE FROM ntf_subscriptions WHERE conn_id = ?" (Only connId)
+      else deleteNtfSubscription' db connId
 
-getNextNtfSubNTFAction :: DB.Connection -> NtfServer -> IO (Either StoreError (Maybe (NtfSubscription, NtfSubNTFAction, NtfActionTs)))
-getNextNtfSubNTFAction db ntfServer@(NtfServer ntfHost ntfPort _) =
-  getWorkItem "ntf NTF" getNtfConnId getNtfSubAction (markNtfSubActionNtfFailed_ db)
+deleteNtfSubscription' :: DB.Connection -> ConnId -> IO ()
+deleteNtfSubscription' db connId = do
+  DB.execute db "DELETE FROM ntf_subscriptions WHERE conn_id = ?" (Only connId)
+
+getNextNtfSubNTFActions :: DB.Connection -> NtfServer -> Int -> IO (Either StoreError [Either StoreError (NtfSubNTFAction, NtfSubscription, NtfActionTs)])
+getNextNtfSubNTFActions db ntfServer@(NtfServer ntfHost ntfPort _) ntfBatchSize =
+  getWorkItems "ntf NTF" getNtfConnIds getNtfSubAction (markNtfSubActionNtfFailed_ db)
   where
-    getNtfConnId :: IO (Maybe ConnId)
-    getNtfConnId =
-      maybeFirstRow fromOnly $
-        DB.query
+    getNtfConnIds :: IO [ConnId]
+    getNtfConnIds =
+      map fromOnly
+        <$> DB.query
           db
           [sql|
             SELECT conn_id
@@ -1597,42 +1720,43 @@ getNextNtfSubNTFAction db ntfServer@(NtfServer ntfHost ntfPort _) =
             WHERE ntf_host = ? AND ntf_port = ? AND ntf_sub_action IS NOT NULL
               AND (ntf_failed = 0 OR updated_by_supervisor = 1)
             ORDER BY ntf_sub_action_ts ASC
-            LIMIT 1
+            LIMIT ?
           |]
-          (ntfHost, ntfPort)
-    getNtfSubAction :: ConnId -> IO (Either StoreError (NtfSubscription, NtfSubNTFAction, NtfActionTs))
+          (ntfHost, ntfPort, ntfBatchSize)
+    getNtfSubAction :: ConnId -> IO (Either StoreError (NtfSubNTFAction, NtfSubscription, NtfActionTs))
     getNtfSubAction connId = do
       markUpdatedByWorker db connId
       firstRow ntfSubAction err $
         DB.query
           db
           [sql|
-            SELECT s.host, s.port, COALESCE(ns.smp_server_key_hash, s.key_hash),
+            SELECT c.user_id, s.host, s.port, COALESCE(ns.smp_server_key_hash, s.key_hash),
               ns.smp_ntf_id, ns.ntf_sub_id, ns.ntf_sub_status, ns.ntf_sub_action_ts, ns.ntf_sub_action
             FROM ntf_subscriptions ns
+            JOIN connections c USING (conn_id)
             JOIN servers s ON s.host = ns.smp_host AND s.port = ns.smp_port
             WHERE ns.conn_id = ?
           |]
           (Only connId)
       where
         err = SEInternal $ "ntf subscription " <> bshow connId <> " returned []"
-        ntfSubAction (smpHost, smpPort, smpKeyHash, ntfQueueId, ntfSubId, ntfSubStatus, actionTs, action) =
+        ntfSubAction (userId, smpHost, smpPort, smpKeyHash, ntfQueueId, ntfSubId, ntfSubStatus, actionTs, action) =
           let smpServer = SMPServer smpHost smpPort smpKeyHash
-              ntfSubscription = NtfSubscription {connId, smpServer, ntfQueueId, ntfServer, ntfSubId, ntfSubStatus}
-           in (ntfSubscription, action, actionTs)
+              ntfSubscription = NtfSubscription {userId, connId, smpServer, ntfQueueId, ntfServer, ntfSubId, ntfSubStatus}
+           in (action, ntfSubscription, actionTs)
 
 markNtfSubActionNtfFailed_ :: DB.Connection -> ConnId -> IO ()
 markNtfSubActionNtfFailed_ db connId =
   DB.execute db "UPDATE ntf_subscriptions SET ntf_failed = 1 where conn_id = ?" (Only connId)
 
-getNextNtfSubSMPAction :: DB.Connection -> SMPServer -> IO (Either StoreError (Maybe (NtfSubscription, NtfSubSMPAction, NtfActionTs)))
-getNextNtfSubSMPAction db smpServer@(SMPServer smpHost smpPort _) =
-  getWorkItem "ntf SMP" getNtfConnId getNtfSubAction (markNtfSubActionSMPFailed_ db)
+getNextNtfSubSMPActions :: DB.Connection -> SMPServer -> Int -> IO (Either StoreError [Either StoreError (NtfSubSMPAction, NtfSubscription)])
+getNextNtfSubSMPActions db smpServer@(SMPServer smpHost smpPort _) ntfBatchSize =
+  getWorkItems "ntf SMP" getNtfConnIds getNtfSubAction (markNtfSubActionSMPFailed_ db)
   where
-    getNtfConnId :: IO (Maybe ConnId)
-    getNtfConnId =
-      maybeFirstRow fromOnly $
-        DB.query
+    getNtfConnIds :: IO [ConnId]
+    getNtfConnIds =
+      map fromOnly
+        <$> DB.query
           db
           [sql|
             SELECT conn_id
@@ -1640,29 +1764,30 @@ getNextNtfSubSMPAction db smpServer@(SMPServer smpHost smpPort _) =
             WHERE smp_host = ? AND smp_port = ? AND ntf_sub_smp_action IS NOT NULL AND ntf_sub_action_ts IS NOT NULL
               AND (smp_failed = 0 OR updated_by_supervisor = 1)
             ORDER BY ntf_sub_action_ts ASC
-            LIMIT 1
+            LIMIT ?
           |]
-          (smpHost, smpPort)
-    getNtfSubAction :: ConnId -> IO (Either StoreError (NtfSubscription, NtfSubSMPAction, NtfActionTs))
+          (smpHost, smpPort, ntfBatchSize)
+    getNtfSubAction :: ConnId -> IO (Either StoreError (NtfSubSMPAction, NtfSubscription))
     getNtfSubAction connId = do
       markUpdatedByWorker db connId
       firstRow ntfSubAction err $
         DB.query
           db
           [sql|
-            SELECT s.ntf_host, s.ntf_port, s.ntf_key_hash,
-              ns.smp_ntf_id, ns.ntf_sub_id, ns.ntf_sub_status, ns.ntf_sub_action_ts, ns.ntf_sub_smp_action
+            SELECT c.user_id, s.ntf_host, s.ntf_port, s.ntf_key_hash,
+              ns.smp_ntf_id, ns.ntf_sub_id, ns.ntf_sub_status, ns.ntf_sub_smp_action
             FROM ntf_subscriptions ns
+            JOIN connections c USING (conn_id)
             JOIN ntf_servers s USING (ntf_host, ntf_port)
             WHERE ns.conn_id = ?
           |]
           (Only connId)
       where
         err = SEInternal $ "ntf subscription " <> bshow connId <> " returned []"
-        ntfSubAction (ntfHost, ntfPort, ntfKeyHash, ntfQueueId, ntfSubId, ntfSubStatus, actionTs, action) =
+        ntfSubAction (userId, ntfHost, ntfPort, ntfKeyHash, ntfQueueId, ntfSubId, ntfSubStatus, action) =
           let ntfServer = NtfServer ntfHost ntfPort ntfKeyHash
-              ntfSubscription = NtfSubscription {connId, smpServer, ntfQueueId, ntfServer, ntfSubId, ntfSubStatus}
-           in (ntfSubscription, action, actionTs)
+              ntfSubscription = NtfSubscription {userId, connId, smpServer, ntfQueueId, ntfServer, ntfSubId, ntfSubStatus}
+           in (action, ntfSubscription)
 
 markNtfSubActionSMPFailed_ :: DB.Connection -> ConnId -> IO ()
 markNtfSubActionSMPFailed_ db connId =
@@ -1693,19 +1818,19 @@ getActiveNtfToken db =
           ntfMode = fromMaybe NMPeriodic ntfMode_
        in NtfToken {deviceToken = DeviceToken provider dt, ntfServer, ntfTokenId, ntfPubKey, ntfPrivKey, ntfDhKeys, ntfDhSecret, ntfTknStatus, ntfTknAction, ntfMode}
 
-getNtfRcvQueue :: DB.Connection -> SMPQueueNtf -> IO (Either StoreError (ConnId, RcvNtfDhSecret))
+getNtfRcvQueue :: DB.Connection -> SMPQueueNtf -> IO (Either StoreError (ConnId, RcvNtfDhSecret, Maybe UTCTime))
 getNtfRcvQueue db SMPQueueNtf {smpServer = (SMPServer host port _), notifierId} =
   firstRow' res SEConnNotFound $
     DB.query
       db
       [sql|
-        SELECT conn_id, rcv_ntf_dh_secret
+        SELECT conn_id, rcv_ntf_dh_secret, last_broker_ts
         FROM rcv_queues
         WHERE host = ? AND port = ? AND ntf_id = ? AND deleted = 0
       |]
       (host, port, notifierId)
   where
-    res (connId, Just rcvNtfDhSecret) = Right (connId, rcvNtfDhSecret)
+    res (connId, Just rcvNtfDhSecret, lastBrokerTs_) = Right (connId, rcvNtfDhSecret, lastBrokerTs_)
     res _ = Left SEConnNotFound
 
 setConnectionNtfs :: DB.Connection -> ConnId -> Bool -> IO ()
@@ -1790,6 +1915,14 @@ instance ToField (Version v) where toField (Version v) = toField v
 
 instance FromField (Version v) where fromField f = Version <$> fromField f
 
+deriving newtype instance ToField EntityId
+
+deriving newtype instance FromField EntityId
+
+deriving newtype instance ToField ChunkReplicaId
+
+deriving newtype instance FromField ChunkReplicaId
+
 listToEither :: e -> [a] -> Either e a
 listToEither _ (x : _) = Right x
 listToEither e _ = Left e
@@ -1906,9 +2039,11 @@ newQueueId_ (Only maxId : _) = DBQueueId (maxId + 1)
 
 getConn :: DB.Connection -> ConnId -> IO (Either StoreError SomeConn)
 getConn = getAnyConn False
+{-# INLINE getConn #-}
 
 getDeletedConn :: DB.Connection -> ConnId -> IO (Either StoreError SomeConn)
 getDeletedConn = getAnyConn True
+{-# INLINE getDeletedConn #-}
 
 getAnyConn :: Bool -> DB.Connection -> ConnId -> IO (Either StoreError SomeConn)
 getAnyConn deleted' dbConn connId =
@@ -1929,9 +2064,11 @@ getAnyConn deleted' dbConn connId =
 
 getConns :: DB.Connection -> [ConnId] -> IO [Either StoreError SomeConn]
 getConns = getAnyConns_ False
+{-# INLINE getConns #-}
 
 getDeletedConns :: DB.Connection -> [ConnId] -> IO [Either StoreError SomeConn]
 getDeletedConns = getAnyConns_ True
+{-# INLINE getDeletedConns #-}
 
 getAnyConns_ :: Bool -> DB.Connection -> [ConnId] -> IO [Either StoreError SomeConn]
 getAnyConns_ deleted' db connIds = forM connIds $ E.handle handleDBError . getAnyConn deleted' db
@@ -1964,6 +2101,10 @@ setConnDeleted db waitDelivery connId
   | otherwise =
       DB.execute db "UPDATE connections SET deleted = ? WHERE conn_id = ?" (True, connId)
 
+setConnUserId :: DB.Connection -> UserId -> ConnId -> UserId -> IO ()
+setConnUserId db oldUserId connId newUserId =
+  DB.execute db "UPDATE connections SET user_id = ? WHERE conn_id = ? and user_id = ?" (newUserId, connId, oldUserId)
+
 setConnAgentVersion :: DB.Connection -> ConnId -> VersionSMPA -> IO ()
 setConnAgentVersion db connId aVersion =
   DB.execute db "UPDATE connections SET smp_agent_version = ? WHERE conn_id = ?" (aVersion, connId)
@@ -2176,9 +2317,9 @@ updateRcvMsgHash db connId sndMsgId internalRcvId internalHash =
 
 -- * updateSndIds helpers
 
-retrieveLastIdsAndHashSnd_ :: DB.Connection -> ConnId -> IO (InternalId, InternalSndId, PrevSndMsgHash)
+retrieveLastIdsAndHashSnd_ :: DB.Connection -> ConnId -> IO (Either StoreError (InternalId, InternalSndId, PrevSndMsgHash))
 retrieveLastIdsAndHashSnd_ dbConn connId = do
-  [(lastInternalId, lastInternalSndId, lastSndHash)] <-
+  firstRow id SEConnNotFound $
     DB.queryNamed
       dbConn
       [sql|
@@ -2187,7 +2328,6 @@ retrieveLastIdsAndHashSnd_ dbConn connId = do
         WHERE conn_id = :conn_id;
       |]
       [":conn_id" := connId]
-  return (lastInternalId, lastInternalSndId, lastSndHash)
 
 updateLastIdsSnd_ :: DB.Connection -> ConnId -> InternalId -> InternalSndId -> IO ()
 updateLastIdsSnd_ dbConn connId newInternalId newInternalSndId =
@@ -2272,8 +2412,8 @@ randomId :: TVar ChaChaDRG -> Int -> IO ByteString
 randomId gVar n = atomically $ U.encode <$> C.randomBytes n gVar
 
 ntfSubAndSMPAction :: NtfSubAction -> (Maybe NtfSubNTFAction, Maybe NtfSubSMPAction)
-ntfSubAndSMPAction (NtfSubNTFAction action) = (Just action, Nothing)
-ntfSubAndSMPAction (NtfSubSMPAction action) = (Nothing, Just action)
+ntfSubAndSMPAction (NSANtf action) = (Just action, Nothing)
+ntfSubAndSMPAction (NSASMP action) = (Nothing, Just action)
 
 createXFTPServer_ :: DB.Connection -> XFTPServer -> IO Int64
 createXFTPServer_ db newSrv@ProtocolServer {host, port, keyHash} =
@@ -2483,7 +2623,7 @@ deleteRcvFile' :: DB.Connection -> DBRcvFileId -> IO ()
 deleteRcvFile' db rcvFileId =
   DB.execute db "DELETE FROM rcv_files WHERE rcv_file_id = ?" (Only rcvFileId)
 
-getNextRcvChunkToDownload :: DB.Connection -> XFTPServer -> NominalDiffTime -> IO (Either StoreError (Maybe (RcvFileChunk, Bool)))
+getNextRcvChunkToDownload :: DB.Connection -> XFTPServer -> NominalDiffTime -> IO (Either StoreError (Maybe (RcvFileChunk, Bool, Maybe RcvFileId)))
 getNextRcvChunkToDownload db server@ProtocolServer {host, port, keyHash} ttl = do
   getWorkItem "rcv_file_download" getReplicaId getChunkData (markRcvFileFailed db . snd)
   where
@@ -2507,7 +2647,7 @@ getNextRcvChunkToDownload db server@ProtocolServer {host, port, keyHash} ttl = d
             LIMIT 1
           |]
           (host, port, keyHash, RFSReceiving, cutoffTs)
-    getChunkData :: (Int64, DBRcvFileId) -> IO (Either StoreError (RcvFileChunk, Bool))
+    getChunkData :: (Int64, DBRcvFileId) -> IO (Either StoreError (RcvFileChunk, Bool, Maybe RcvFileId))
     getChunkData (rcvFileChunkReplicaId, _fileId) =
       firstRow toChunk SEFileNotFound $
         DB.query
@@ -2516,7 +2656,7 @@ getNextRcvChunkToDownload db server@ProtocolServer {host, port, keyHash} ttl = d
             SELECT
               f.rcv_file_id, f.rcv_file_entity_id, f.user_id, c.rcv_file_chunk_id, c.chunk_no, c.chunk_size, c.digest, f.tmp_path, c.tmp_path,
               r.rcv_file_chunk_replica_id, r.replica_id, r.replica_key, r.received, r.delay, r.retries,
-              f.approved_relays
+              f.approved_relays, f.redirect_entity_id
             FROM rcv_file_chunk_replicas r
             JOIN xftp_servers s ON s.xftp_server_id = r.xftp_server_id
             JOIN rcv_file_chunks c ON c.rcv_file_chunk_id = r.rcv_file_chunk_id
@@ -2525,8 +2665,8 @@ getNextRcvChunkToDownload db server@ProtocolServer {host, port, keyHash} ttl = d
           |]
           (Only rcvFileChunkReplicaId)
       where
-        toChunk :: ((DBRcvFileId, RcvFileId, UserId, Int64, Int, FileSize Word32, FileDigest, FilePath, Maybe FilePath) :. (Int64, ChunkReplicaId, C.APrivateAuthKey, Bool, Maybe Int64, Int) :. Only Bool) -> (RcvFileChunk, Bool)
-        toChunk ((rcvFileId, rcvFileEntityId, userId, rcvChunkId, chunkNo, chunkSize, digest, fileTmpPath, chunkTmpPath) :. (rcvChunkReplicaId, replicaId, replicaKey, received, delay, retries) :. (Only approvedRelays)) =
+        toChunk :: ((DBRcvFileId, RcvFileId, UserId, Int64, Int, FileSize Word32, FileDigest, FilePath, Maybe FilePath) :. (Int64, ChunkReplicaId, C.APrivateAuthKey, Bool, Maybe Int64, Int) :. (Bool, Maybe RcvFileId)) -> (RcvFileChunk, Bool, Maybe RcvFileId)
+        toChunk ((rcvFileId, rcvFileEntityId, userId, rcvChunkId, chunkNo, chunkSize, digest, fileTmpPath, chunkTmpPath) :. (rcvChunkReplicaId, replicaId, replicaKey, received, delay, retries) :. (approvedRelays, redirectEntityId_)) =
           ( RcvFileChunk
               { rcvFileId,
                 rcvFileEntityId,
@@ -2539,7 +2679,8 @@ getNextRcvChunkToDownload db server@ProtocolServer {host, port, keyHash} ttl = d
                 chunkTmpPath,
                 replicas = [RcvFileChunkReplica {rcvChunkReplicaId, server, replicaId, replicaKey, received, delay, retries}]
               },
-            approvedRelays
+            approvedRelays,
+            redirectEntityId_
           )
 
 getNextRcvFileToDecrypt :: DB.Connection -> NominalDiffTime -> IO (Either StoreError (Maybe RcvFile))
@@ -3041,10 +3182,9 @@ getServersStats db =
   firstRow id SEServersStatsNotFound $
     DB.query_ db "SELECT started_at, servers_stats FROM servers_stats WHERE servers_stats_id = 1"
 
-resetServersStats :: DB.Connection -> IO ()
-resetServersStats db = do
-  currentTs <- getCurrentTime
-  DB.execute db "UPDATE servers_stats SET servers_stats = NULL, started_at = ?, updated_at = ? WHERE servers_stats_id = 1" (currentTs, currentTs)
+resetServersStats :: DB.Connection -> UTCTime -> IO ()
+resetServersStats db startedAt =
+  DB.execute db "UPDATE servers_stats SET servers_stats = NULL, started_at = ?, updated_at = ? WHERE servers_stats_id = 1" (startedAt, startedAt)
 
 $(J.deriveJSON defaultJSON ''UpMigration)
 

src/Simplex/Messaging/Agent/Store/SQLite/Common.hs

@@ -1,4 +1,5 @@
 {-# LANGUAGE DuplicateRecordFields #-}
+{-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE NamedFieldPuns #-}
 {-# LANGUAGE ScopedTypeVariables #-}
 
@@ -8,20 +9,20 @@ module Simplex.Messaging.Agent.Store.SQLite.Common
     withConnection',
     withTransaction,
     withTransaction',
-    withTransactionCtx,
+    withTransactionPriority,
     dbBusyLoop,
     storeKey,
   )
 where
 
 import Control.Concurrent (threadDelay)
+import Control.Concurrent.STM (retry)
 import Data.ByteArray (ScrubbedBytes)
 import qualified Data.ByteArray as BA
-import Data.Time.Clock (diffUTCTime, getCurrentTime)
 import Database.SQLite.Simple (SQLError)
 import qualified Database.SQLite.Simple as SQL
 import qualified Simplex.Messaging.Agent.Store.SQLite.DB as DB
-import Simplex.Messaging.Util (diffToMilliseconds)
+import Simplex.Messaging.Util (ifM, unlessM)
 import qualified UnliftIO.Exception as E
 import UnliftIO.MVar
 import UnliftIO.STM
@@ -32,35 +33,40 @@ storeKey key keepKey = if keepKey || BA.null key then Just key else Nothing
 data SQLiteStore = SQLiteStore
   { dbFilePath :: FilePath,
     dbKey :: TVar (Maybe ScrubbedBytes),
+    dbSem :: TVar Int,
     dbConnection :: MVar DB.Connection,
     dbClosed :: TVar Bool,
     dbNew :: Bool
   }
 
+withConnectionPriority :: SQLiteStore -> Bool -> (DB.Connection -> IO a) -> IO a
+withConnectionPriority SQLiteStore {dbSem, dbConnection} priority action
+  | priority = E.bracket_ signal release $ withMVar dbConnection action
+  | otherwise = lowPriority
+  where
+    lowPriority = wait >> withMVar dbConnection (\db -> ifM free (Just <$> action db) (pure Nothing)) >>= maybe lowPriority pure
+    signal = atomically $ modifyTVar' dbSem (+ 1)
+    release = atomically $ modifyTVar' dbSem $ \sem -> if sem > 0 then sem - 1 else 0
+    wait = unlessM free $ atomically $ unlessM ((0 ==) <$> readTVar dbSem) retry
+    free = (0 ==) <$> readTVarIO dbSem
+
 withConnection :: SQLiteStore -> (DB.Connection -> IO a) -> IO a
-withConnection SQLiteStore {dbConnection} = withMVar dbConnection
+withConnection st = withConnectionPriority st False
 
 withConnection' :: SQLiteStore -> (SQL.Connection -> IO a) -> IO a
 withConnection' st action = withConnection st $ action . DB.conn
 
-withTransaction :: SQLiteStore -> (DB.Connection -> IO a) -> IO a
-withTransaction = withTransactionCtx Nothing
-
 withTransaction' :: SQLiteStore -> (SQL.Connection -> IO a) -> IO a
 withTransaction' st action = withTransaction st $ action . DB.conn
 
-withTransactionCtx :: Maybe String -> SQLiteStore -> (DB.Connection -> IO a) -> IO a
-withTransactionCtx ctx_ st action = withConnection st $ dbBusyLoop . transactionWithCtx
+withTransaction :: SQLiteStore -> (DB.Connection -> IO a) -> IO a
+withTransaction st = withTransactionPriority st False
+{-# INLINE withTransaction #-}
+
+withTransactionPriority :: SQLiteStore -> Bool -> (DB.Connection -> IO a) -> IO a
+withTransactionPriority st priority action = withConnectionPriority st priority $ dbBusyLoop . transaction
   where
-    transactionWithCtx db@DB.Connection {conn} = case ctx_ of
-      Nothing -> SQL.withImmediateTransaction conn $ action db
-      Just ctx -> do
-        t1 <- getCurrentTime
-        r <- SQL.withImmediateTransaction conn $ action db
-        t2 <- getCurrentTime
-        putStrLn $ "withTransactionCtx start :: " <> show t1 <> " :: " <> ctx
-        putStrLn $ "withTransactionCtx end   :: " <> show t2 <> " :: " <> ctx <> " :: duration=" <> show (diffToMilliseconds $ diffUTCTime t2 t1)
-        pure r
+    transaction db@DB.Connection {conn} = SQL.withImmediateTransaction conn $ action db
 
 dbBusyLoop :: forall a. IO a -> IO a
 dbBusyLoop action = loop 500 3000000

src/Simplex/Messaging/Agent/Store/SQLite/DB.hs

@@ -1,4 +1,5 @@
 {-# LANGUAGE DeriveAnyClass #-}
+{-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE NamedFieldPuns #-}
 {-# LANGUAGE StrictData #-}
 {-# LANGUAGE TemplateHaskell #-}
@@ -20,15 +21,19 @@ where
 
 import Control.Concurrent.STM
 import Control.Monad (when)
+import Control.Exception
 import qualified Data.Aeson.TH as J
 import Data.Int (Int64)
+import Data.Map.Strict (Map)
+import qualified Data.Map.Strict as M
+import Data.Text (Text)
 import Data.Time (diffUTCTime, getCurrentTime)
 import Database.SQLite.Simple (FromRow, NamedParam, Query, ToRow)
 import qualified Database.SQLite.Simple as SQL
 import Simplex.Messaging.Parsers (defaultJSON)
 import Simplex.Messaging.TMap (TMap)
 import qualified Simplex.Messaging.TMap as TM
-import Simplex.Messaging.Util (diffToMilliseconds)
+import Simplex.Messaging.Util (diffToMilliseconds, tshow)
 
 data Connection = Connection
   { conn :: SQL.Connection,
@@ -38,33 +43,41 @@ data Connection = Connection
 data SlowQueryStats = SlowQueryStats
   { count :: Int64,
     timeMax :: Int64,
-    timeAvg :: Int64
+    timeAvg :: Int64,
+    errs :: Map Text Int
   }
   deriving (Show)
 
 timeIt :: TMap Query SlowQueryStats -> Query -> IO a -> IO a
 timeIt slow sql a = do
   t <- getCurrentTime
-  r <- a
+  r <- a `catch` \e -> do
+    atomically $ TM.alter (Just . updateQueryErrors e) sql slow
+    throwIO e
   t' <- getCurrentTime
   let diff = diffToMilliseconds $ diffUTCTime t' t
-  atomically $ when (diff > 5) $ TM.alter (updateQueryStats diff) sql slow
+  when (diff > 1) $ atomically $ TM.alter (updateQueryStats diff) sql slow
   pure r
   where
+    updateQueryErrors :: SomeException -> Maybe SlowQueryStats -> SlowQueryStats
+    updateQueryErrors e Nothing = SlowQueryStats 0 0 0 $ M.singleton (tshow e) 1
+    updateQueryErrors e (Just stats@SlowQueryStats {errs}) =
+      stats {errs = M.alter (Just . maybe 1 (+ 1)) (tshow e) errs}
     updateQueryStats :: Int64 -> Maybe SlowQueryStats -> Maybe SlowQueryStats
-    updateQueryStats diff Nothing = Just $ SlowQueryStats 1 diff diff
-    updateQueryStats diff (Just SlowQueryStats {count, timeMax, timeAvg}) =
+    updateQueryStats diff Nothing = Just $ SlowQueryStats 1 diff diff M.empty
+    updateQueryStats diff (Just SlowQueryStats {count, timeMax, timeAvg, errs}) =
       Just $
         SlowQueryStats
           { count = count + 1,
             timeMax = max timeMax diff,
-            timeAvg = (timeAvg * count + diff) `div` (count + 1)
+            timeAvg = (timeAvg * count + diff) `div` (count + 1),
+            errs
           }
 
 open :: String -> IO Connection
 open f = do
   conn <- SQL.open f
-  slow <- atomically $ TM.empty
+  slow <- TM.emptyIO
   pure Connection {conn, slow}
 
 close :: Connection -> IO ()

src/Simplex/Messaging/Agent/Store/SQLite/Migrations.hs

@@ -29,7 +29,7 @@ import Control.Monad (forM_, when)
 import qualified Data.Aeson.TH as J
 import Data.List (intercalate, sortOn)
 import Data.List.NonEmpty (NonEmpty)
-import qualified Data.Map as M
+import qualified Data.Map.Strict as M
 import Data.Maybe (isNothing, mapMaybe)
 import Data.Text (Text)
 import Data.Text.Encoding (decodeLatin1)
@@ -73,6 +73,9 @@ import Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240223_connections_wai
 import Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240225_ratchet_kem
 import Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240417_rcv_files_approved_relays
 import Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240624_snd_secure
+import Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240702_servers_stats
+import Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240930_ntf_tokens_to_delete
+import Simplex.Messaging.Agent.Store.SQLite.Migrations.M20241007_rcv_queues_last_broker_ts
 import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Parsers (dropPrefix, sumTypeJSON)
 import Simplex.Messaging.Transport.Client (TransportHost)
@@ -114,7 +117,10 @@ schemaMigrations =
     ("m20240223_connections_wait_delivery", m20240223_connections_wait_delivery, Just down_m20240223_connections_wait_delivery),
     ("m20240225_ratchet_kem", m20240225_ratchet_kem, Just down_m20240225_ratchet_kem),
     ("m20240417_rcv_files_approved_relays", m20240417_rcv_files_approved_relays, Just down_m20240417_rcv_files_approved_relays),
-    ("m20240624_snd_secure", m20240624_snd_secure, Just down_m20240624_snd_secure)
+    ("m20240624_snd_secure", m20240624_snd_secure, Just down_m20240624_snd_secure),
+    ("m20240702_servers_stats", m20240702_servers_stats, Just down_m20240702_servers_stats),
+    ("m20240930_ntf_tokens_to_delete", m20240930_ntf_tokens_to_delete, Just down_m20240930_ntf_tokens_to_delete),
+    ("m20241007_rcv_queues_last_broker_ts", m20241007_rcv_queues_last_broker_ts, Just down_m20241007_rcv_queues_last_broker_ts)
   ]
 
 -- | The list of migrations in ascending order by date

src/Simplex/Messaging/Agent/Store/SQLite/Migrations/M20240702_servers_stats.hs

@@ -1,6 +1,6 @@
 {-# LANGUAGE QuasiQuotes #-}
 
-module Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240518_servers_stats where
+module Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240702_servers_stats where
 
 import Database.SQLite.Simple (Query)
 import Database.SQLite.Simple.QQ (sql)
@@ -8,8 +8,8 @@ import Database.SQLite.Simple.QQ (sql)
 -- servers_stats_id: dummy id, there should always only be one record with servers_stats_id = 1
 -- servers_stats: overall accumulated stats, past and session, reset to null on stats reset
 -- started_at: starting point of tracking stats, reset on stats reset
-m20240518_servers_stats :: Query
-m20240518_servers_stats =
+m20240702_servers_stats :: Query
+m20240702_servers_stats =
     [sql|
 CREATE TABLE servers_stats(
   servers_stats_id INTEGER PRIMARY KEY,
@@ -22,8 +22,8 @@ CREATE TABLE servers_stats(
 INSERT INTO servers_stats (servers_stats_id) VALUES (1);
 |]
 
-down_m20240518_servers_stats :: Query
-down_m20240518_servers_stats =
+down_m20240702_servers_stats :: Query
+down_m20240702_servers_stats =
     [sql|
 DROP TABLE servers_stats;
 |]

src/Simplex/Messaging/Agent/Store/SQLite/Migrations/M20240930_ntf_tokens_to_delete.hs

@@ -0,0 +1,27 @@
+{-# LANGUAGE QuasiQuotes #-}
+
+module Simplex.Messaging.Agent.Store.SQLite.Migrations.M20240930_ntf_tokens_to_delete where
+
+import Database.SQLite.Simple (Query)
+import Database.SQLite.Simple.QQ (sql)
+
+m20240930_ntf_tokens_to_delete :: Query
+m20240930_ntf_tokens_to_delete =
+    [sql|
+CREATE TABLE ntf_tokens_to_delete (
+  ntf_token_to_delete_id INTEGER PRIMARY KEY,
+  ntf_host TEXT NOT NULL,
+  ntf_port TEXT NOT NULL,
+  ntf_key_hash BLOB NOT NULL,
+  tkn_id BLOB NOT NULL, -- token ID assigned by notifications server
+  tkn_priv_key BLOB NOT NULL, -- client's private key to sign token commands,
+  del_failed INTEGER DEFAULT 0,
+  created_at TEXT NOT NULL DEFAULT (datetime('now'))
+);
+|]
+
+down_m20240930_ntf_tokens_to_delete :: Query
+down_m20240930_ntf_tokens_to_delete =
+    [sql|
+DROP TABLE ntf_tokens_to_delete;
+|]

src/Simplex/Messaging/Agent/Store/SQLite/Migrations/M20241007_rcv_queues_last_broker_ts.hs

@@ -0,0 +1,18 @@
+{-# LANGUAGE QuasiQuotes #-}
+
+module Simplex.Messaging.Agent.Store.SQLite.Migrations.M20241007_rcv_queues_last_broker_ts where
+
+import Database.SQLite.Simple (Query)
+import Database.SQLite.Simple.QQ (sql)
+
+m20241007_rcv_queues_last_broker_ts :: Query
+m20241007_rcv_queues_last_broker_ts =
+    [sql|
+ALTER TABLE rcv_queues ADD COLUMN last_broker_ts TEXT;
+|]
+
+down_m20241007_rcv_queues_last_broker_ts :: Query
+down_m20241007_rcv_queues_last_broker_ts =
+    [sql|
+ALTER TABLE rcv_queues DROP COLUMN last_broker_ts;
+|]

src/Simplex/Messaging/Agent/Store/SQLite/Migrations/agent_lint.sql

@@ -0,0 +1 @@
+CREATE INDEX 'ratchets_conn_id' ON 'ratchets'('conn_id'); --> connections(conn_id)

src/Simplex/Messaging/Agent/Store/SQLite/Migrations/agent_schema.sql

@@ -56,6 +56,7 @@ CREATE TABLE rcv_queues(
   switch_status TEXT,
   deleted INTEGER NOT NULL DEFAULT 0,
   snd_secure INTEGER NOT NULL DEFAULT 0,
+  last_broker_ts TEXT,
   PRIMARY KEY(host, port, rcv_id),
   FOREIGN KEY(host, port) REFERENCES servers
   ON DELETE RESTRICT ON UPDATE CASCADE,
@@ -396,6 +397,23 @@ CREATE TABLE processed_ratchet_key_hashes(
   created_at TEXT NOT NULL DEFAULT(datetime('now')),
   updated_at TEXT NOT NULL DEFAULT(datetime('now'))
 );
+CREATE TABLE servers_stats(
+  servers_stats_id INTEGER PRIMARY KEY,
+  servers_stats TEXT,
+  started_at TEXT NOT NULL DEFAULT(datetime('now')),
+  created_at TEXT NOT NULL DEFAULT(datetime('now')),
+  updated_at TEXT NOT NULL DEFAULT(datetime('now'))
+);
+CREATE TABLE ntf_tokens_to_delete(
+  ntf_token_to_delete_id INTEGER PRIMARY KEY,
+  ntf_host TEXT NOT NULL,
+  ntf_port TEXT NOT NULL,
+  ntf_key_hash BLOB NOT NULL,
+  tkn_id BLOB NOT NULL, -- token ID assigned by notifications server
+  tkn_priv_key BLOB NOT NULL, -- client's private key to sign token commands,
+del_failed INTEGER DEFAULT 0,
+created_at TEXT NOT NULL DEFAULT(datetime('now'))
+);
 CREATE UNIQUE INDEX idx_rcv_queues_ntf ON rcv_queues(host, port, ntf_id);
 CREATE UNIQUE INDEX idx_rcv_queue_id ON rcv_queues(conn_id, rcv_queue_id);
 CREATE UNIQUE INDEX idx_snd_queue_id ON snd_queues(conn_id, snd_queue_id);

src/Simplex/Messaging/Agent/TRcvQueues.hs

@@ -1,7 +1,9 @@
+{-# LANGUAGE FlexibleInstances #-}
 {-# LANGUAGE LambdaCase #-}
 
 module Simplex.Messaging.Agent.TRcvQueues
   ( TRcvQueues (getRcvQueues, getConnections),
+    Queue (..),
     empty,
     clear,
     deleteConn,
@@ -9,9 +11,9 @@ module Simplex.Messaging.Agent.TRcvQueues
     addQueue,
     batchAddQueues,
     deleteQueue,
+    hasSessQueues,
     getSessQueues,
     getDelSessQueues,
-    qKey,
   )
 where
 
@@ -25,46 +27,51 @@ import Simplex.Messaging.Agent.Store (RcvQueue, StoredRcvQueue (..))
 import Simplex.Messaging.Protocol (RecipientId, SMPServer)
 import Simplex.Messaging.TMap (TMap)
 import qualified Simplex.Messaging.TMap as TM
+import Simplex.Messaging.Transport
+
+class Queue q where
+  connId' :: q -> ConnId
+  qKey :: q -> (UserId, SMPServer, RecipientId)
 
 -- the fields in this record have the same data with swapped keys for lookup efficiency,
 -- and all methods must maintain this invariant.
-data TRcvQueues = TRcvQueues
-  { getRcvQueues :: TMap (UserId, SMPServer, RecipientId) RcvQueue,
+data TRcvQueues q = TRcvQueues
+  { getRcvQueues :: TMap (UserId, SMPServer, RecipientId) q,
     getConnections :: TMap ConnId (NonEmpty (UserId, SMPServer, RecipientId))
   }
 
-empty :: STM TRcvQueues
-empty = TRcvQueues <$> TM.empty <*> TM.empty
+empty :: IO (TRcvQueues q)
+empty = TRcvQueues <$> TM.emptyIO <*> TM.emptyIO
 
-clear :: TRcvQueues -> STM ()
+clear :: TRcvQueues q -> STM ()
 clear (TRcvQueues qs cs) = TM.clear qs >> TM.clear cs
 
-deleteConn :: ConnId -> TRcvQueues -> STM ()
+deleteConn :: ConnId -> TRcvQueues q -> STM ()
 deleteConn cId (TRcvQueues qs cs) =
   TM.lookupDelete cId cs >>= \case
     Just ks -> modifyTVar' qs $ \qs' -> foldl' (flip M.delete) qs' ks
     Nothing -> pure ()
 
-hasConn :: ConnId -> TRcvQueues -> STM Bool
+hasConn :: ConnId -> TRcvQueues q -> STM Bool
 hasConn cId (TRcvQueues _ cs) = TM.member cId cs
 
-addQueue :: RcvQueue -> TRcvQueues -> STM ()
+addQueue :: Queue q => q -> TRcvQueues q -> STM ()
 addQueue rq (TRcvQueues qs cs) = do
   TM.insert k rq qs
-  TM.alter addQ (connId rq) cs
+  TM.alter addQ (connId' rq) cs
   where
     addQ = Just . maybe (k :| []) (k <|)
     k = qKey rq
 
--- Save time by aggregating modifyTVar
-batchAddQueues :: Foldable t => TRcvQueues -> t RcvQueue -> STM ()
+-- Save time by aggregating modifyTVar'
+batchAddQueues :: (Foldable t, Queue q) => TRcvQueues q -> t q -> STM ()
 batchAddQueues (TRcvQueues qs cs) rqs = do
   modifyTVar' qs $ \now -> foldl' (\rqs' rq -> M.insert (qKey rq) rq rqs') now rqs
-  modifyTVar' cs $ \now -> foldl' (\cs' rq -> M.alter (addQ $ qKey rq) (connId rq) cs') now rqs
+  modifyTVar' cs $ \now -> foldl' (\cs' rq -> M.alter (addQ $ qKey rq) (connId' rq) cs') now rqs
   where
     addQ k = Just . maybe (k :| []) (k <|)
 
-deleteQueue :: RcvQueue -> TRcvQueues -> STM ()
+deleteQueue :: RcvQueue -> TRcvQueues RcvQueue -> STM ()
 deleteQueue rq (TRcvQueues qs cs) = do
   TM.delete k qs
   TM.update delQ (connId rq) cs
@@ -72,21 +79,25 @@ deleteQueue rq (TRcvQueues qs cs) = do
     delQ = L.nonEmpty . L.filter (/= k)
     k = qKey rq
 
-getSessQueues :: (UserId, SMPServer, Maybe ConnId) -> TRcvQueues -> STM [RcvQueue]
-getSessQueues tSess (TRcvQueues qs _) = M.foldl' addQ [] <$> readTVar qs
+hasSessQueues :: (UserId, SMPServer, Maybe ConnId) -> TRcvQueues RcvQueue -> STM Bool
+hasSessQueues tSess (TRcvQueues qs _) = any (`isSession` tSess) <$> readTVar qs
+
+getSessQueues :: (UserId, SMPServer, Maybe ConnId) -> TRcvQueues RcvQueue -> IO [RcvQueue]
+getSessQueues tSess (TRcvQueues qs _) = M.foldl' addQ [] <$> readTVarIO qs
   where
     addQ qs' rq = if rq `isSession` tSess then rq : qs' else qs'
 
-getDelSessQueues :: (UserId, SMPServer, Maybe ConnId) -> TRcvQueues -> STM ([RcvQueue], [ConnId])
-getDelSessQueues tSess (TRcvQueues qs cs) = do
+getDelSessQueues :: (UserId, SMPServer, Maybe ConnId) -> SessionId -> TRcvQueues (SessionId, RcvQueue) -> STM ([RcvQueue], [ConnId])
+getDelSessQueues tSess sessId' (TRcvQueues qs cs) = do
   (removedQs, qs'') <- (\qs' -> M.foldl' delQ ([], qs') qs') <$> readTVar qs
   writeTVar qs $! qs''
   removedConns <- stateTVar cs $ \cs' -> foldl' delConn ([], cs') removedQs
   pure (removedQs, removedConns)
   where
-    delQ acc@(removed, qs') rq
-      | rq `isSession` tSess = (rq : removed, M.delete (qKey rq) qs')
+    delQ acc@(removed, qs') (sessId, rq)
+      | rq `isSession` tSess && sessId == sessId' = (rq : removed, M.delete (qKey rq) qs')
       | otherwise = acc
+    delConn :: ([ConnId], M.Map ConnId (NonEmpty (UserId, SMPServer, RecipientId))) -> RcvQueue -> ([ConnId], M.Map ConnId (NonEmpty (UserId, SMPServer, RecipientId)))
     delConn (removed, cs') rq = M.alterF f cId cs'
       where
         cId = connId rq
@@ -100,5 +111,10 @@ isSession :: RcvQueue -> (UserId, SMPServer, Maybe ConnId) -> Bool
 isSession rq (uId, srv, connId_) =
   userId rq == uId && server rq == srv && maybe True (connId rq ==) connId_
 
-qKey :: RcvQueue -> (UserId, SMPServer, ConnId)
-qKey rq = (userId rq, server rq, connId rq)
+instance Queue RcvQueue where
+  connId' = connId
+  qKey rq = (userId rq, server rq, rcvId rq)
+
+instance Queue (SessionId, RcvQueue) where
+  connId' = connId . snd
+  qKey = qKey . snd

src/Simplex/Messaging/Client.hs

@@ -63,11 +63,13 @@ module Simplex.Messaging.Client
     forwardSMPTransmission,
     getSMPQueueInfo,
     sendProtocolCommand,
+    sendProtocolCommands,
 
     -- * Supporting types and client configuration
     ProtocolClientError (..),
     SMPClientError,
     ProxyClientError (..),
+    Response (..),
     unexpectedResponse,
     ProtocolClientConfig (..),
     NetworkConfig (..),
@@ -80,10 +82,11 @@ module Simplex.Messaging.Client
     defaultSMPClientConfig,
     defaultNetworkConfig,
     transportClientConfig,
+    clientSocksCredentials,
     chooseTransportHost,
-    proxyUsername,
     temporaryClientError,
     smpProxyError,
+    textToHostMode,
     ServerTransmissionBatch,
     ServerTransmission (..),
     ClientCommand,
@@ -116,29 +119,32 @@ import qualified Data.Aeson.TH as J
 import qualified Data.Attoparsec.ByteString.Char8 as A
 import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Char8 as B
+import qualified Data.ByteString.Base64 as B64
 import Data.Functor (($>))
 import Data.Int (Int64)
 import Data.List (find)
 import Data.List.NonEmpty (NonEmpty (..))
 import qualified Data.List.NonEmpty as L
 import Data.Maybe (catMaybes, fromMaybe)
+import Data.Text (Text)
+import qualified Data.Text as T
 import Data.Time.Clock (UTCTime (..), diffUTCTime, getCurrentTime)
 import qualified Data.X509 as X
 import qualified Data.X509.Validation as XV
 import Network.Socket (ServiceName)
+import Network.Socks5 (SocksCredentials (..))
 import Numeric.Natural
 import qualified Simplex.Messaging.Crypto as C
 import Simplex.Messaging.Encoding
 import Simplex.Messaging.Encoding.String
-import Simplex.Messaging.Parsers (defaultJSON, dropPrefix, enumJSON)
+import Simplex.Messaging.Parsers (defaultJSON, dropPrefix, enumJSON, sumTypeJSON)
 import Simplex.Messaging.Protocol
 import Simplex.Messaging.Server.QueueStore.QueueInfo
 import Simplex.Messaging.TMap (TMap)
 import qualified Simplex.Messaging.TMap as TM
 import Simplex.Messaging.Transport
-import Simplex.Messaging.Transport.Client (SocksProxy, TransportClientConfig (..), TransportHost (..), defaultTcpConnectTimeout, runTransportClient)
+import Simplex.Messaging.Transport.Client (SocksAuth (..), SocksProxyWithAuth (..), TransportClientConfig (..), TransportHost (..), defaultSMPPort, defaultTcpConnectTimeout, runTransportClient)
 import Simplex.Messaging.Transport.KeepAlive
-import Simplex.Messaging.Transport.WebSockets (WS)
 import Simplex.Messaging.Util (bshow, diffToMicroseconds, ifM, liftEitherWith, raceAny_, threadDelay', tshow, whenM)
 import Simplex.Messaging.Version
 import System.Mem.Weak (Weak, deRefWeak)
@@ -170,17 +176,17 @@ data PClient v err msg = PClient
     msgQ :: Maybe (TBQueue (ServerTransmissionBatch v err msg))
   }
 
-smpClientStub :: TVar ChaChaDRG -> ByteString -> VersionSMP -> Maybe (THandleAuth 'TClient) -> STM SMPClient
+smpClientStub :: TVar ChaChaDRG -> ByteString -> VersionSMP -> Maybe (THandleAuth 'TClient) -> IO SMPClient
 smpClientStub g sessionId thVersion thAuth = do
   let ts = UTCTime (read "2024-03-31") 0
-  connected <- newTVar False
-  clientCorrId <- C.newRandomDRG g
-  sentCommands <- TM.empty
-  sendPings <- newTVar False
-  lastReceived <- newTVar ts
-  timeoutErrorCount <- newTVar 0
-  sndQ <- newTBQueue 100
-  rcvQ <- newTBQueue 100
+  connected <- newTVarIO False
+  clientCorrId <- atomically $ C.newRandomDRG g
+  sentCommands <- TM.emptyIO
+  sendPings <- newTVarIO False
+  lastReceived <- newTVarIO ts
+  timeoutErrorCount <- newTVarIO 0
+  sndQ <- newTBQueueIO 100
+  rcvQ <- newTBQueueIO 100
   return
     ProtocolClient
       { action = Nothing,
@@ -192,6 +198,7 @@ smpClientStub g sessionId thVersion thAuth = do
               thAuth,
               blockSize = smpBlockSize,
               implySessId = thVersion >= authCmdsSMPVersion,
+              encryptBlock = Nothing,
               batch = True
             },
         sessionTs = ts,
@@ -236,18 +243,34 @@ data HostMode
     HMPublic
   deriving (Eq, Show)
 
+textToHostMode :: Text -> Either String HostMode
+textToHostMode = \case
+  "public" -> Right HMPublic
+  "onion" -> Right HMOnionViaSocks
+  s -> Left $ T.unpack $ "Invalid host_mode: " <> s
+
 data SocksMode
   = -- | always use SOCKS proxy when enabled
     SMAlways
   | -- | use SOCKS proxy only for .onion hosts when no public host is available
-    -- This mode is used in SMP proxy to minimize SOCKS proxy usage.
+    -- This mode is used in SMP proxy and in notifications server to minimize SOCKS proxy usage.
     SMOnion
   deriving (Eq, Show)
 
+instance StrEncoding SocksMode where
+  strEncode = \case
+    SMAlways -> "always"
+    SMOnion -> "onion"
+  strP =
+    A.takeTill (== ' ') >>= \case
+      "always" -> pure SMAlways
+      "onion" -> pure SMOnion
+      _ -> fail "Invalid Socks mode"
+
 -- | network configuration for the client
 data NetworkConfig = NetworkConfig
   { -- | use SOCKS5 proxy
-    socksProxy :: Maybe SocksProxy,
+    socksProxy :: Maybe SocksProxyWithAuth,
     -- | when to use SOCKS proxy
     socksMode :: SocksMode,
     -- | determines critera which host is chosen from the list
@@ -260,6 +283,8 @@ data NetworkConfig = NetworkConfig
     smpProxyMode :: SMPProxyMode,
     -- | Fallback to direct connection when destination SMP relay does not support SMP proxy protocol extensions
     smpProxyFallback :: SMPProxyFallback,
+    -- | use web port 443 for SMP protocol
+    smpWebPort :: Bool,
     -- | timeout for the initial client TCP/TLS connection (microseconds)
     tcpConnectTimeout :: Int,
     -- | timeout of protocol commands (microseconds)
@@ -278,7 +303,7 @@ data NetworkConfig = NetworkConfig
   }
   deriving (Eq, Show)
 
-data TransportSessionMode = TSMUser | TSMEntity
+data TransportSessionMode = TSMUser | TSMSession | TSMServer | TSMEntity
   deriving (Eq, Show)
 
 -- SMP proxy mode for sending messages
@@ -328,9 +353,10 @@ defaultNetworkConfig =
       socksMode = SMAlways,
       hostMode = HMOnionViaSocks,
       requiredHostMode = False,
-      sessionMode = TSMUser,
+      sessionMode = TSMSession,
       smpProxyMode = SPMNever,
       smpProxyFallback = SPFAllow,
+      smpWebPort = False,
       tcpConnectTimeout = defaultTcpConnectTimeout,
       tcpTimeout = 15_000_000,
       tcpTimeoutPerKb = 5_000,
@@ -341,15 +367,31 @@ defaultNetworkConfig =
       logTLSErrors = False
     }
 
-transportClientConfig :: NetworkConfig -> TransportHost -> TransportClientConfig
-transportClientConfig NetworkConfig {socksProxy, socksMode, tcpConnectTimeout, tcpKeepAlive, logTLSErrors} host =
-  TransportClientConfig {socksProxy = useSocksProxy socksMode, tcpConnectTimeout, tcpKeepAlive, logTLSErrors, clientCredentials = Nothing, alpn = Nothing}
+transportClientConfig :: NetworkConfig -> TransportHost -> Bool -> TransportClientConfig
+transportClientConfig NetworkConfig {socksProxy, socksMode, tcpConnectTimeout, tcpKeepAlive, logTLSErrors} host useSNI =
+  TransportClientConfig {socksProxy = useSocksProxy socksMode, tcpConnectTimeout, tcpKeepAlive, logTLSErrors, clientCredentials = Nothing, alpn = Nothing, useSNI}
   where
-    useSocksProxy SMAlways = socksProxy
+    socksProxy' = (\(SocksProxyWithAuth _ proxy) -> proxy) <$> socksProxy
+    useSocksProxy SMAlways = socksProxy'
     useSocksProxy SMOnion = case host of
-      THOnionHost _ -> socksProxy
+      THOnionHost _ -> socksProxy'
       _ -> Nothing
-{-# INLINE transportClientConfig #-}
+
+clientSocksCredentials :: ProtocolTypeI (ProtoType msg) => NetworkConfig -> UTCTime -> TransportSession msg -> Maybe SocksCredentials
+clientSocksCredentials NetworkConfig {socksProxy, sessionMode} proxySessTs (userId, srv, entityId_) = case socksProxy of
+  Just (SocksProxyWithAuth auth _) -> case auth of
+    SocksAuthUsername {username, password} -> Just $ SocksCredentials username password
+    SocksAuthNull -> Nothing
+    SocksIsolateByAuth -> Just $ SocksCredentials sessionUsername ""
+  Nothing -> Nothing
+  where
+    sessionUsername =
+      B64.encode $ C.sha256Hash $
+        bshow userId <> case sessionMode of
+          TSMUser -> ""
+          TSMSession -> ":" <> bshow proxySessTs
+          TSMServer -> ":" <> bshow proxySessTs <> "@" <> strEncode srv
+          TSMEntity -> ":" <> bshow proxySessTs <> "@" <> strEncode srv <> maybe "" ("/" <>) entityId_
 
 -- | protocol client configuration.
 data ProtocolClientConfig v = ProtocolClientConfig
@@ -363,24 +405,31 @@ data ProtocolClientConfig v = ProtocolClientConfig
     -- | client-server protocol version range
     serverVRange :: VersionRange v,
     -- | agree shared session secret (used in SMP proxy for additional encryption layer)
-    agreeSecret :: Bool
+    agreeSecret :: Bool,
+    -- | send SNI to server, False for SMP
+    useSNI :: Bool
   }
 
 -- | Default protocol client configuration.
-defaultClientConfig :: Maybe [ALPN] -> VersionRange v -> ProtocolClientConfig v
-defaultClientConfig clientALPN serverVRange =
+defaultClientConfig :: Maybe [ALPN] -> Bool -> VersionRange v -> ProtocolClientConfig v
+defaultClientConfig clientALPN useSNI serverVRange =
   ProtocolClientConfig
     { qSize = 64,
       defaultTransport = ("443", transport @TLS),
       networkConfig = defaultNetworkConfig,
       clientALPN,
       serverVRange,
-      agreeSecret = False
+      agreeSecret = False,
+      useSNI
     }
 {-# INLINE defaultClientConfig #-}
 
 defaultSMPClientConfig :: ProtocolClientConfig SMPVersion
-defaultSMPClientConfig = defaultClientConfig (Just supportedSMPHandshakes) supportedClientSMPRelayVRange
+defaultSMPClientConfig =
+  (defaultClientConfig (Just supportedSMPHandshakes) False supportedClientSMPRelayVRange)
+    { defaultTransport = (show defaultSMPPort, transport @TLS),
+      agreeSecret = True
+    }
 {-# INLINE defaultSMPClientConfig #-}
 
 data Request err msg = Request
@@ -431,32 +480,33 @@ transportSession' = transportSession . client_
 type UserId = Int64
 
 -- | Transport session key - includes entity ID if `sessionMode = TSMEntity`.
-type TransportSession msg = (UserId, ProtoServer msg, Maybe EntityId)
+-- Please note that for SMP connection ID is used as entity ID, not queue ID.
+type TransportSession msg = (UserId, ProtoServer msg, Maybe ByteString)
 
 -- | Connects to 'ProtocolServer' using passed client configuration
 -- and queue for messages and notifications.
 --
 -- A single queue can be used for multiple 'SMPClient' instances,
 -- as 'SMPServerTransmission' includes server information.
-getProtocolClient :: forall v err msg. Protocol v err msg => TVar ChaChaDRG -> TransportSession msg -> ProtocolClientConfig v -> Maybe (TBQueue (ServerTransmissionBatch v err msg)) -> (ProtocolClient v err msg -> IO ()) -> IO (Either (ProtocolClientError err) (ProtocolClient v err msg))
-getProtocolClient g transportSession@(_, srv, _) cfg@ProtocolClientConfig {qSize, networkConfig, clientALPN, serverVRange, agreeSecret} msgQ disconnected = do
+getProtocolClient :: forall v err msg. Protocol v err msg => TVar ChaChaDRG -> TransportSession msg -> ProtocolClientConfig v -> Maybe (TBQueue (ServerTransmissionBatch v err msg)) -> UTCTime -> (ProtocolClient v err msg -> IO ()) -> IO (Either (ProtocolClientError err) (ProtocolClient v err msg))
+getProtocolClient g transportSession@(_, srv, _) cfg@ProtocolClientConfig {qSize, networkConfig, clientALPN, serverVRange, agreeSecret, useSNI} msgQ proxySessTs disconnected = do
   case chooseTransportHost networkConfig (host srv) of
     Right useHost ->
-      (getCurrentTime >>= atomically . mkProtocolClient useHost >>= runClient useTransport useHost)
+      (getCurrentTime >>= mkProtocolClient useHost >>= runClient useTransport useHost)
         `catch` \(e :: IOException) -> pure . Left $ PCEIOError e
     Left e -> pure $ Left e
   where
-    NetworkConfig {tcpConnectTimeout, tcpTimeout, smpPingInterval} = networkConfig
-    mkProtocolClient :: TransportHost -> UTCTime -> STM (PClient v err msg)
+    NetworkConfig {smpWebPort, tcpConnectTimeout, tcpTimeout, smpPingInterval} = networkConfig
+    mkProtocolClient :: TransportHost -> UTCTime -> IO (PClient v err msg)
     mkProtocolClient transportHost ts = do
-      connected <- newTVar False
-      sendPings <- newTVar False
-      lastReceived <- newTVar ts
-      timeoutErrorCount <- newTVar 0
-      clientCorrId <- C.newRandomDRG g
-      sentCommands <- TM.empty
-      sndQ <- newTBQueue qSize
-      rcvQ <- newTBQueue qSize
+      connected <- newTVarIO False
+      sendPings <- newTVarIO False
+      lastReceived <- newTVarIO ts
+      timeoutErrorCount <- newTVarIO 0
+      clientCorrId <- atomically $ C.newRandomDRG g
+      sentCommands <- TM.emptyIO
+      sndQ <- newTBQueueIO qSize
+      rcvQ <- newTBQueueIO qSize
       return
         PClient
           { connected,
@@ -477,10 +527,10 @@ getProtocolClient g transportSession@(_, srv, _) cfg@ProtocolClientConfig {qSize
     runClient :: (ServiceName, ATransport) -> TransportHost -> PClient v err msg -> IO (Either (ProtocolClientError err) (ProtocolClient v err msg))
     runClient (port', ATransport t) useHost c = do
       cVar <- newEmptyTMVarIO
-      let tcConfig = (transportClientConfig networkConfig useHost) {alpn = clientALPN}
-          username = proxyUsername transportSession
+      let tcConfig = (transportClientConfig networkConfig useHost useSNI) {alpn = clientALPN}
+          socksCreds = clientSocksCredentials networkConfig proxySessTs transportSession
       tId <-
-        runTransportClient tcConfig (Just username) useHost port' (Just $ keyHash srv) (client t c cVar)
+        runTransportClient tcConfig socksCreds useHost port' (Just $ keyHash srv) (client t c cVar)
           `forkFinally` \_ -> void (atomically . tryPutTMVar cVar $ Left PCENetworkError)
       c_ <- tcpConnectTimeout `timeout` atomically (takeTMVar cVar)
       case c_ of
@@ -490,8 +540,9 @@ getProtocolClient g transportSession@(_, srv, _) cfg@ProtocolClientConfig {qSize
 
     useTransport :: (ServiceName, ATransport)
     useTransport = case port srv of
-      "" -> defaultTransport cfg
-      "80" -> ("80", transport @WS)
+      "" -> case protocolTypeI @(ProtoType msg) of
+        SPSMP | smpWebPort -> ("443", transport @TLS)
+        _ -> defaultTransport cfg
       p -> (p, transport @TLS)
 
     client :: forall c. Transport c => TProxy c -> PClient v err msg -> TMVar (Either (ProtocolClientError err) (ProtocolClient v err msg)) -> c -> IO ()
@@ -534,7 +585,7 @@ getProtocolClient g transportSession@(_, srv, _) cfg@ProtocolClientConfig {qSize
           if remaining > 1_000_000 -- delay pings only for significant time
             then loop remaining
             else do
-              whenM (readTVarIO sendPings) $ void . runExceptT $ sendProtocolCommand c Nothing "" (protocolPing @v @err @msg)
+              whenM (readTVarIO sendPings) $ void . runExceptT $ sendProtocolCommand c Nothing NoEntity (protocolPing @v @err @msg)
               -- sendProtocolCommand/getResponse updates counter for each command
               cnt <- readTVarIO timeoutErrorCount
               -- drop client when maxCnt of commands have timed out in sequence, but only after some time has passed after last received response
@@ -555,7 +606,7 @@ getProtocolClient g transportSession@(_, srv, _) cfg@ProtocolClientConfig {qSize
     processMsg ProtocolClient {client_ = PClient {sentCommands}} (_, _, (corrId, entId, respOrErr))
       | B.null $ bs corrId = sendMsg $ STEvent clientResp
       | otherwise =
-          atomically (TM.lookup corrId sentCommands) >>= \case
+          TM.lookupIO corrId sentCommands >>= \case
             Nothing -> sendMsg $ STUnexpectedError unexpected
             Just Request {entityId, command, pending, responseVar} -> do
               wasPending <-
@@ -586,10 +637,6 @@ getProtocolClient g transportSession@(_, srv, _) cfg@ProtocolClientConfig {qSize
 unexpectedResponse :: Show r => r -> ProtocolClientError err
 unexpectedResponse = PCEUnexpectedResponse . B.pack . take 32 . show
 
-proxyUsername :: TransportSession msg -> ByteString
-proxyUsername (userId, _, entityId_) = C.sha256Hash $ bshow userId <> maybe "" (":" <>) entityId_
-{-# INLINE proxyUsername #-}
-
 -- | Disconnects client from the server and terminates client threads.
 closeProtocolClient :: ProtocolClient v err msg -> IO ()
 closeProtocolClient = mapM_ (deRefWeak >=> mapM_ killThread) . action
@@ -660,7 +707,7 @@ createSMPQueue ::
   Bool ->
   ExceptT SMPClientError IO QueueIdsKeys
 createSMPQueue c (rKey, rpKey) dhKey auth subMode sndSecure =
-  sendSMPCommand c (Just rpKey) "" (NEW rKey dhKey auth subMode sndSecure) >>= \case
+  sendSMPCommand c (Just rpKey) NoEntity (NEW rKey dhKey auth subMode sndSecure) >>= \case
     IDS qik -> pure qik
     r -> throwE $ unexpectedResponse r
 
@@ -668,8 +715,8 @@ createSMPQueue c (rKey, rpKey) dhKey auth subMode sndSecure =
 --
 -- https://github.com/simplex-chat/simplexmq/blob/master/protocol/simplex-messaging.md#subscribe-to-queue
 subscribeSMPQueue :: SMPClient -> RcvPrivateAuthKey -> RecipientId -> ExceptT SMPClientError IO ()
-subscribeSMPQueue c@ProtocolClient {client_ = PClient {sendPings}} rpKey rId = do
-  liftIO . atomically $ writeTVar sendPings True
+subscribeSMPQueue c rpKey rId = do
+  liftIO $ enablePings c
   sendSMPCommand c (Just rpKey) rId SUB >>= \case
     OK -> pure ()
     cmd@MSG {} -> liftIO $ writeSMPMessage c rId cmd
@@ -677,8 +724,8 @@ subscribeSMPQueue c@ProtocolClient {client_ = PClient {sendPings}} rpKey rId = d
 
 -- | Subscribe to multiple SMP queues batching commands if supported.
 subscribeSMPQueues :: SMPClient -> NonEmpty (RcvPrivateAuthKey, RecipientId) -> IO (NonEmpty (Either SMPClientError ()))
-subscribeSMPQueues c@ProtocolClient {client_ = PClient {sendPings}} qs = do
-  atomically $ writeTVar sendPings True
+subscribeSMPQueues c qs = do
+  liftIO $ enablePings c
   sendProtocolCommands c cs >>= mapM (processSUBResponse c)
   where
     cs = L.map (\(rpKey, rId) -> (Just rpKey, rId, Cmd SRecipient SUB)) qs
@@ -717,14 +764,22 @@ getSMPMessage c rpKey rId =
 --
 -- https://github.com/simplex-chat/simplexmq/blob/master/protocol/simplex-messaging.md#subscribe-to-queue-notifications
 subscribeSMPQueueNotifications :: SMPClient -> NtfPrivateAuthKey -> NotifierId -> ExceptT SMPClientError IO ()
-subscribeSMPQueueNotifications = okSMPCommand NSUB
+subscribeSMPQueueNotifications c npKey nId = do
+  liftIO $ enablePings c
+  okSMPCommand NSUB c npKey nId
 {-# INLINE subscribeSMPQueueNotifications #-}
 
 -- | Subscribe to multiple SMP queues notifications batching commands if supported.
 subscribeSMPQueuesNtfs :: SMPClient -> NonEmpty (NtfPrivateAuthKey, NotifierId) -> IO (NonEmpty (Either SMPClientError ()))
-subscribeSMPQueuesNtfs = okSMPCommands NSUB
+subscribeSMPQueuesNtfs c qs = do
+  liftIO $ enablePings c
+  okSMPCommands NSUB c qs
 {-# INLINE subscribeSMPQueuesNtfs #-}
 
+enablePings :: SMPClient -> IO ()
+enablePings ProtocolClient {client_ = PClient {sendPings}} = atomically $ writeTVar sendPings True
+{-# INLINE enablePings #-}
+
 -- | Secure the SMP queue by adding a sender public key.
 --
 -- https://github.com/simplex-chat/simplexmq/blob/master/protocol/simplex-messaging.md#secure-queue-command
@@ -819,11 +874,11 @@ deleteSMPQueues = okSMPCommands DEL
 connectSMPProxiedRelay :: SMPClient -> SMPServer -> Maybe BasicAuth -> ExceptT SMPClientError IO ProxiedRelay
 connectSMPProxiedRelay c@ProtocolClient {client_ = PClient {tcpConnectTimeout, tcpTimeout}} relayServ@ProtocolServer {keyHash = C.KeyHash kh} proxyAuth
   | thVersion (thParams c) >= sendingProxySMPVersion =
-      sendProtocolCommand_ c Nothing tOut Nothing "" (Cmd SProxiedClient (PRXY relayServ proxyAuth)) >>= \case
+      sendProtocolCommand_ c Nothing tOut Nothing NoEntity (Cmd SProxiedClient (PRXY relayServ proxyAuth)) >>= \case
         PKEY sId vr (chain, key) ->
           case supportedClientSMPRelayVRange `compatibleVersion` vr of
             Nothing -> throwE $ transportErr TEVersion
-            Just (Compatible v) -> liftEitherWith (const $ transportErr $ TEHandshake IDENTITY) $ ProxiedRelay sId v <$> validateRelay chain key
+            Just (Compatible v) -> liftEitherWith (const $ transportErr $ TEHandshake IDENTITY) $ ProxiedRelay sId v proxyAuth <$> validateRelay chain key
         r -> throwE $ unexpectedResponse r
   | otherwise = throwE $ PCETransportError TEVersion
   where
@@ -842,16 +897,17 @@ connectSMPProxiedRelay c@ProtocolClient {client_ = PClient {tcpConnectTimeout, t
 data ProxiedRelay = ProxiedRelay
   { prSessionId :: SessionId,
     prVersion :: VersionSMP,
+    prBasicAuth :: Maybe BasicAuth, -- auth is included here to allow reconnecting via the same proxy after NO_SESSION error
     prServerKey :: C.PublicKeyX25519
   }
 
 data ProxyClientError
   = -- | protocol error response from proxy
-    ProxyProtocolError ErrorType
+    ProxyProtocolError {protocolErr :: ErrorType}
   | -- | unexpexted response
-    ProxyUnexpectedResponse String
+    ProxyUnexpectedResponse {responseStr :: String}
   | -- | error between proxy and server
-    ProxyResponseError ErrorType
+    ProxyResponseError {responseErr :: ErrorType}
   deriving (Eq, Show, Exception)
 
 instance StrEncoding ProxyClientError where
@@ -902,7 +958,7 @@ proxySMPCommand ::
   SenderId ->
   Command 'Sender ->
   ExceptT SMPClientError IO (Either ProxyClientError ())
-proxySMPCommand c@ProtocolClient {thParams = proxyThParams, client_ = PClient {clientCorrId = g, tcpTimeout}} (ProxiedRelay sessionId v serverKey) spKey sId command = do
+proxySMPCommand c@ProtocolClient {thParams = proxyThParams, client_ = PClient {clientCorrId = g, tcpTimeout}} (ProxiedRelay sessionId v _ serverKey) spKey sId command = do
   -- prepare params
   let serverThAuth = (\ta -> ta {serverPeerPubKey = serverKey}) <$> thAuth proxyThParams
       serverThParams = smpTHParamsSetVersion v proxyThParams {sessionId, thAuth = serverThAuth}
@@ -920,7 +976,7 @@ proxySMPCommand c@ProtocolClient {thParams = proxyThParams, client_ = PClient {c
   et <- liftEitherWith PCECryptoError $ EncTransmission <$> C.cbEncrypt cmdSecret nonce b paddedProxiedTLength
   -- proxy interaction errors are wrapped
   let tOut = Just $ 2 * tcpTimeout
-  tryE (sendProtocolCommand_ c (Just nonce) tOut Nothing sessionId (Cmd SProxiedClient (PFWD v cmdPubKey et))) >>= \case
+  tryE (sendProtocolCommand_ c (Just nonce) tOut Nothing (EntityId sessionId) (Cmd SProxiedClient (PFWD v cmdPubKey et))) >>= \case
     Right r -> case r of
       PRES (EncResponse er) -> do
         -- server interaction errors are thrown directly
@@ -956,7 +1012,7 @@ forwardSMPTransmission c@ProtocolClient {thParams, client_ = PClient {clientCorr
   let fwdT = FwdTransmission {fwdCorrId, fwdVersion, fwdKey, fwdTransmission}
       eft = EncFwdTransmission $ C.cbEncryptNoPad sessSecret nonce (smpEncode fwdT)
   -- send
-  sendProtocolCommand_ c (Just nonce) Nothing Nothing "" (Cmd SSender (RFWD eft)) >>= \case
+  sendProtocolCommand_ c (Just nonce) Nothing Nothing NoEntity (Cmd SSender (RFWD eft)) >>= \case
     RRES (EncFwdResponse efr) -> do
       -- unwrap
       r' <- liftEitherWith PCECryptoError $ C.cbDecryptNoPad sessSecret (C.reverseNonce nonce) efr
@@ -1004,7 +1060,7 @@ sendProtocolCommands c@ProtocolClient {thParams = THandleParams {batch, blockSiz
       | diff == 0 = pure $ L.fromList rs
       | diff > 0 = do
           putStrLn "send error: fewer responses than expected"
-          pure $ L.fromList $ rs <> replicate diff (Response "" $ Left $ PCETransportError TEBadBlock)
+          pure $ L.fromList $ rs <> replicate diff (Response NoEntity $ Left $ PCETransportError TEBadBlock)
       | otherwise = do
           putStrLn "send error: more responses than expected"
           pure $ L.fromList $ take (L.length cs) rs
@@ -1078,13 +1134,13 @@ mkTransmission_ ProtocolClient {thParams, client_ = PClient {clientCorrId, sentC
   nonce@(C.CbNonce corrId) <- maybe (atomically $ C.randomCbNonce clientCorrId) pure nonce_
   let TransmissionForAuth {tForAuth, tToSend} = encodeTransmissionForAuth thParams (CorrId corrId, entityId, command)
       auth = authTransmission (thAuth thParams) pKey_ nonce tForAuth
-  r <- atomically $ mkRequest (CorrId corrId)
+  r <- mkRequest (CorrId corrId)
   pure ((,tToSend) <$> auth, r)
   where
-    mkRequest :: CorrId -> STM (Request err msg)
+    mkRequest :: CorrId -> IO (Request err msg)
     mkRequest corrId = do
-      pending <- newTVar True
-      responseVar <- newEmptyTMVar
+      pending <- newTVarIO True
+      responseVar <- newEmptyTMVarIO
       let r =
             Request
               { corrId,
@@ -1093,7 +1149,7 @@ mkTransmission_ ProtocolClient {thParams, client_ = PClient {clientCorrId, sentC
                 pending,
                 responseVar
               }
-      TM.insert corrId r sentCommands
+      atomically $ TM.insert corrId r sentCommands
       pure r
 
 authTransmission :: Maybe (THandleAuth 'TClient) -> Maybe C.APrivateAuthKey -> C.CbNonce -> ByteString -> Either TransportError (Maybe TransmissionAuth)
@@ -1139,6 +1195,6 @@ $(J.deriveJSON (enumJSON $ dropPrefix "SPF") ''SMPProxyFallback)
 
 $(J.deriveJSON defaultJSON ''NetworkConfig)
 
-$(J.deriveJSON (enumJSON $ dropPrefix "Proxy") ''ProxyClientError)
+$(J.deriveJSON (sumTypeJSON $ dropPrefix "Proxy") ''ProxyClientError)
 
 $(J.deriveJSON defaultJSON ''TBQueueInfo)

src/Simplex/Messaging/Client/Agent.hs

@@ -20,17 +20,15 @@ import Control.Monad.Except
 import Control.Monad.IO.Unlift
 import Control.Monad.Trans.Except
 import Crypto.Random (ChaChaDRG)
-import Data.Bifunctor (bimap, first)
+import Data.Bifunctor (first)
 import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Char8 as B
-import Data.Either (partitionEithers)
-import Data.List (partition)
 import Data.List.NonEmpty (NonEmpty)
 import qualified Data.List.NonEmpty as L
 import Data.Map.Strict (Map)
 import qualified Data.Map.Strict as M
-import Data.Maybe (listToMaybe)
 import Data.Set (Set)
+import qualified Data.Set as S
 import Data.Text.Encoding
 import Data.Time.Clock (NominalDiffTime, UTCTime, addUTCTime, getCurrentTime)
 import Data.Tuple (swap)
@@ -55,8 +53,8 @@ type SMPClientVar = SessionVar (Either (SMPClientError, Maybe UTCTime) (OwnServe
 data SMPClientAgentEvent
   = CAConnected SMPServer
   | CADisconnected SMPServer (Set SMPSub)
-  | CAResubscribed SMPServer (NonEmpty SMPSub)
-  | CASubError SMPServer (NonEmpty (SMPSub, SMPClientError))
+  | CASubscribed SMPServer SMPSubParty (NonEmpty QueueId)
+  | CASubError SMPServer SMPSubParty (NonEmpty (QueueId, SMPClientError))
 
 data SMPSubParty = SPRecipient | SPNotifier
   deriving (Eq, Ord, Show)
@@ -78,7 +76,7 @@ data SMPClientAgentConfig = SMPClientAgentConfig
 defaultSMPClientAgentConfig :: SMPClientAgentConfig
 defaultSMPClientAgentConfig =
   SMPClientAgentConfig
-    { smpCfg = defaultSMPClientConfig {defaultTransport = ("5223", transport @TLS)},
+    { smpCfg = defaultSMPClientConfig,
       reconnectInterval =
         RetryInterval
           { initialInterval = second,
@@ -86,9 +84,9 @@ defaultSMPClientAgentConfig =
             maxInterval = 10 * second
           },
       persistErrorInterval = 30, -- seconds
-      msgQSize = 256,
-      agentQSize = 256,
-      agentSubsBatchSize = 900,
+      msgQSize = 2048,
+      agentQSize = 2048,
+      agentSubsBatchSize = 1360,
       ownServerDomains = []
     }
   where
@@ -97,12 +95,13 @@ defaultSMPClientAgentConfig =
 data SMPClientAgent = SMPClientAgent
   { agentCfg :: SMPClientAgentConfig,
     active :: TVar Bool,
+    startedAt :: UTCTime,
     msgQ :: TBQueue (ServerTransmissionBatch SMPVersion ErrorType BrokerMsg),
     agentQ :: TBQueue SMPClientAgentEvent,
     randomDrg :: TVar ChaChaDRG,
     smpClients :: TMap SMPServer SMPClientVar,
     smpSessions :: TMap SessionId (OwnServer, SMPClient),
-    srvSubs :: TMap SMPServer (TMap SMPSub C.APrivateAuthKey),
+    srvSubs :: TMap SMPServer (TMap SMPSub (SessionId, C.APrivateAuthKey)),
     pendingSrvSubs :: TMap SMPServer (TMap SMPSub C.APrivateAuthKey),
     smpSubWorkers :: TMap SMPServer (SessionVar (Async ())),
     workerSeq :: TVar Int
@@ -110,21 +109,23 @@ data SMPClientAgent = SMPClientAgent
 
 type OwnServer = Bool
 
-newSMPClientAgent :: SMPClientAgentConfig -> TVar ChaChaDRG -> STM SMPClientAgent
+newSMPClientAgent :: SMPClientAgentConfig -> TVar ChaChaDRG -> IO SMPClientAgent
 newSMPClientAgent agentCfg@SMPClientAgentConfig {msgQSize, agentQSize} randomDrg = do
-  active <- newTVar True
-  msgQ <- newTBQueue msgQSize
-  agentQ <- newTBQueue agentQSize
-  smpClients <- TM.empty
-  smpSessions <- TM.empty
-  srvSubs <- TM.empty
-  pendingSrvSubs <- TM.empty
-  smpSubWorkers <- TM.empty
-  workerSeq <- newTVar 0
+  active <- newTVarIO True
+  startedAt <- getCurrentTime
+  msgQ <- newTBQueueIO msgQSize
+  agentQ <- newTBQueueIO agentQSize
+  smpClients <- TM.emptyIO
+  smpSessions <- TM.emptyIO
+  srvSubs <- TM.emptyIO
+  pendingSrvSubs <- TM.emptyIO
+  smpSubWorkers <- TM.emptyIO
+  workerSeq <- newTVarIO 0
   pure
     SMPClientAgent
       { agentCfg,
         active,
+        startedAt,
         msgQ,
         agentQ,
         randomDrg,
@@ -192,12 +193,12 @@ getSMPServerClient'' ca@SMPClientAgent {agentCfg, smpClients, smpSessions, worke
 isOwnServer :: SMPClientAgent -> SMPServer -> OwnServer
 isOwnServer SMPClientAgent {agentCfg} ProtocolServer {host} =
   let srv = strEncode $ L.head host
-   in any (\s -> s == srv || (B.cons '.' s) `B.isSuffixOf` srv) (ownServerDomains agentCfg)
+   in any (\s -> s == srv || B.cons '.' s `B.isSuffixOf` srv) (ownServerDomains agentCfg)
 
 -- | Run an SMP client for SMPClientVar
 connectClient :: SMPClientAgent -> SMPServer -> SMPClientVar -> IO (Either SMPClientError SMPClient)
-connectClient ca@SMPClientAgent {agentCfg, smpClients, smpSessions, msgQ, randomDrg} srv v =
-  getProtocolClient randomDrg (1, srv, Nothing) (smpCfg agentCfg) (Just msgQ) clientDisconnected
+connectClient ca@SMPClientAgent {agentCfg, smpClients, smpSessions, msgQ, randomDrg, startedAt} srv v =
+  getProtocolClient randomDrg (1, srv, Nothing) (smpCfg agentCfg) (Just msgQ) startedAt clientDisconnected
   where
     clientDisconnected :: SMPClient -> IO ()
     clientDisconnected smp = do
@@ -206,20 +207,17 @@ connectClient ca@SMPClientAgent {agentCfg, smpClients, smpSessions, msgQ, random
 
     removeClientAndSubs :: SMPClient -> IO (Maybe (Map SMPSub C.APrivateAuthKey))
     removeClientAndSubs smp = atomically $ do
+      TM.delete sessId smpSessions
       removeSessVar v srv smpClients
-      TM.delete (sessionId $ thParams smp) smpSessions
-      TM.lookupDelete srv (srvSubs ca) >>= mapM updateSubs
+      TM.lookup srv (srvSubs ca) >>= mapM updateSubs
       where
+        sessId = sessionId $ thParams smp
         updateSubs sVar = do
-          ss <- readTVar sVar
-          addPendingSubs sVar ss
-          pure ss
-
-        addPendingSubs sVar ss = do
-          let ps = pendingSrvSubs ca
-          TM.lookup srv ps >>= \case
-            Just ss' -> TM.union ss ss'
-            _ -> TM.insert srv sVar ps
+          -- removing subscriptions that have matching sessionId to disconnected client
+          -- and keep the other ones (they can be made by the new client)
+          pending <- M.map snd <$> stateTVar sVar (M.partition ((sessId ==) . fst))
+          addSubs_ (pendingSrvSubs ca) srv pending
+          pure pending
 
     serverDown :: Map SMPSub C.APrivateAuthKey -> IO ()
     serverDown ss = unless (M.null ss) $ do
@@ -234,7 +232,7 @@ reconnectClient ca@SMPClientAgent {active, agentCfg, smpSubWorkers, workerSeq} s
   where
     getWorkerVar ts =
       ifM
-        (null <$> getPending)
+        (noPending)
         (pure Nothing) -- prevent race with cleanup and adding pending queues in another call
         (Just <$> getSessVar workerSeq srv smpSubWorkers ts)
     newSubWorker :: SessionVar (Async ()) -> IO ()
@@ -243,12 +241,13 @@ reconnectClient ca@SMPClientAgent {active, agentCfg, smpSubWorkers, workerSeq} s
       atomically $ putTMVar (sessionVar v) a
     runSubWorker =
       withRetryInterval (reconnectInterval agentCfg) $ \_ loop -> do
-        pending <- atomically getPending
-        forM_ pending $ \cs -> whenM (readTVarIO active) $ do
-          void $ tcpConnectTimeout `timeout` runExceptT (reconnectSMPClient ca srv cs)
+        pending <- liftIO getPending
+        unless (null pending) $ whenM (readTVarIO active) $ do
+          void $ tcpConnectTimeout `timeout` runExceptT (reconnectSMPClient ca srv pending)
           loop
     ProtocolClientConfig {networkConfig = NetworkConfig {tcpConnectTimeout}} = smpCfg agentCfg
-    getPending = mapM readTVar =<< TM.lookup srv (pendingSrvSubs ca)
+    noPending = maybe (pure True) (fmap M.null . readTVar) =<< TM.lookup srv (pendingSrvSubs ca)
+    getPending = maybe (pure M.empty) readTVarIO =<< TM.lookupIO srv (pendingSrvSubs ca)
     cleanup :: SessionVar (Async ()) -> STM ()
     cleanup v = do
       -- Here we wait until TMVar is not empty to prevent worker cleanup happening before worker is added to TMVar.
@@ -258,32 +257,22 @@ reconnectClient ca@SMPClientAgent {active, agentCfg, smpSubWorkers, workerSeq} s
 
 reconnectSMPClient :: SMPClientAgent -> SMPServer -> Map SMPSub C.APrivateAuthKey -> ExceptT SMPClientError IO ()
 reconnectSMPClient ca@SMPClientAgent {agentCfg} srv cs =
-  withSMP ca srv $ \smp -> do
-    subs' <- filterM (fmap not . atomically . hasSub (srvSubs ca) srv . fst) $ M.assocs cs
-    let (nSubs, rSubs) = partition (isNotifier . fst . fst) subs'
+  withSMP ca srv $ \smp -> liftIO $ do
+    currSubs <- maybe (pure M.empty) readTVarIO =<< TM.lookupIO srv (srvSubs ca)
+    let (nSubs, rSubs) = foldr (groupSub currSubs) ([], []) $ M.assocs cs
     subscribe_ smp SPNotifier nSubs
     subscribe_ smp SPRecipient rSubs
   where
-    isNotifier = \case
-      SPNotifier -> True
-      SPRecipient -> False
-    subscribe_ :: SMPClient -> SMPSubParty -> [(SMPSub, C.APrivateAuthKey)] -> ExceptT SMPClientError IO ()
-    subscribe_ smp party = mapM_ subscribeBatch . toChunks (agentSubsBatchSize agentCfg)
+    groupSub :: Map SMPSub (SessionId, C.APrivateAuthKey) -> (SMPSub, C.APrivateAuthKey) -> ([(QueueId, C.APrivateAuthKey)], [(QueueId, C.APrivateAuthKey)]) -> ([(QueueId, C.APrivateAuthKey)], [(QueueId, C.APrivateAuthKey)])
+    groupSub currSubs (s@(party, qId), k) acc@(nSubs, rSubs)
+      | M.member s currSubs = acc
+      | otherwise = case party of
+          SPNotifier -> (s' : nSubs, rSubs)
+          SPRecipient -> (nSubs, s' : rSubs)
       where
-        subscribeBatch subs' = do
-          let subs'' :: (NonEmpty (QueueId, C.APrivateAuthKey)) = L.map (first snd) subs'
-          rs <- liftIO $ smpSubscribeQueues party ca smp srv subs''
-          let rs' :: (NonEmpty ((SMPSub, C.APrivateAuthKey), Either SMPClientError ())) =
-                L.zipWith (first . const) subs' rs
-              rs'' :: [Either (SMPSub, SMPClientError) (SMPSub, C.APrivateAuthKey)] =
-                map (\(sub, r) -> bimap (fst sub,) (const sub) r) $ L.toList rs'
-              (errs, oks) = partitionEithers rs''
-              (tempErrs, finalErrs) = partition (temporaryClientError . snd) errs
-          mapM_ (atomically . addSubscription ca srv) oks
-          mapM_ (notify ca . CAResubscribed srv) $ L.nonEmpty $ map fst oks
-          mapM_ (atomically . removePendingSubscription ca srv . fst) finalErrs
-          mapM_ (notify ca . CASubError srv) $ L.nonEmpty finalErrs
-          mapM_ (throwE . snd) $ listToMaybe tempErrs
+        s' = (qId, k)
+    subscribe_ :: SMPClient -> SMPSubParty -> [(QueueId, C.APrivateAuthKey)] -> IO ()
+    subscribe_ smp party = mapM_ (smpSubscribeQueues party ca smp srv) . toChunks (agentSubsBatchSize agentCfg)
 
 notify :: MonadIO m => SMPClientAgent -> SMPClientAgentEvent -> m ()
 notify ca evt = atomically $ writeTBQueue (agentQ ca) evt
@@ -297,14 +286,15 @@ getConnectedSMPServerClient SMPClientAgent {smpClients} srv =
     $>>= \case
       (_, Right r) -> pure $ Just $ Right r
       (v, Left (e, ts_)) ->
-        pure ts_ $>>= \ts ->  -- proxy will create a new connection if ts_ is Nothing
+        pure ts_ $>>= \ts ->
+          -- proxy will create a new connection if ts_ is Nothing
           ifM
             ((ts <) <$> liftIO getCurrentTime) -- error persistence interval period expired?
             (Nothing <$ atomically (removeSessVar v srv smpClients)) -- proxy will create a new connection
             (pure $ Just $ Left e) -- not expired, returning error
 
-lookupSMPServerClient :: SMPClientAgent -> SessionId -> STM (Maybe (OwnServer, SMPClient))
-lookupSMPServerClient SMPClientAgent {smpSessions} sessId = TM.lookup sessId smpSessions
+lookupSMPServerClient :: SMPClientAgent -> SessionId -> IO (Maybe (OwnServer, SMPClient))
+lookupSMPServerClient SMPClientAgent {smpSessions} sessId = TM.lookupIO sessId smpSessions
 
 closeSMPClientAgent :: SMPClientAgent -> IO ()
 closeSMPClientAgent c = do
@@ -334,86 +324,102 @@ withSMP ca srv action = (getSMPServerClient' ca srv >>= action) `catchE` logSMPE
       liftIO $ putStrLn $ "SMP error (" <> show srv <> "): " <> show e
       throwE e
 
-subscribeQueue :: SMPClientAgent -> SMPServer -> (SMPSub, C.APrivateAuthKey) -> ExceptT SMPClientError IO ()
-subscribeQueue ca srv sub = do
-  atomically $ addPendingSubscription ca srv sub
-  withSMP ca srv $ \smp -> subscribe_ smp `catchE` handleErr
-  where
-    subscribe_ smp = do
-      smpSubscribe smp sub
-      atomically $ addSubscription ca srv sub
-
-    handleErr e = do
-      atomically . when (e /= PCENetworkError && e /= PCEResponseTimeout) $
-        removePendingSubscription ca srv (fst sub)
-      throwE e
-
-subscribeQueuesSMP :: SMPClientAgent -> SMPServer -> NonEmpty (RecipientId, RcvPrivateAuthKey) -> IO (NonEmpty (RecipientId, Either SMPClientError ()))
+subscribeQueuesSMP :: SMPClientAgent -> SMPServer -> NonEmpty (RecipientId, RcvPrivateAuthKey) -> IO ()
 subscribeQueuesSMP = subscribeQueues_ SPRecipient
 
-subscribeQueuesNtfs :: SMPClientAgent -> SMPServer -> NonEmpty (NotifierId, NtfPrivateAuthKey) -> IO (NonEmpty (NotifierId, Either SMPClientError ()))
+subscribeQueuesNtfs :: SMPClientAgent -> SMPServer -> NonEmpty (NotifierId, NtfPrivateAuthKey) -> IO ()
 subscribeQueuesNtfs = subscribeQueues_ SPNotifier
 
-subscribeQueues_ :: SMPSubParty -> SMPClientAgent -> SMPServer -> NonEmpty (QueueId, C.APrivateAuthKey) -> IO (NonEmpty (QueueId, Either SMPClientError ()))
+subscribeQueues_ :: SMPSubParty -> SMPClientAgent -> SMPServer -> NonEmpty (QueueId, C.APrivateAuthKey) -> IO ()
 subscribeQueues_ party ca srv subs = do
-  atomically $ forM_ subs $ addPendingSubscription ca srv . first (party,)
+  atomically $ addPendingSubs ca srv party $ L.toList subs
   runExceptT (getSMPServerClient' ca srv) >>= \case
-    Left e -> pure $ L.map ((,Left e) . fst) subs
     Right smp -> smpSubscribeQueues party ca smp srv subs
+    Left _ -> pure () -- no call to reconnectClient - failing getSMPServerClient' does that
 
-smpSubscribeQueues :: SMPSubParty -> SMPClientAgent -> SMPClient -> SMPServer -> NonEmpty (QueueId, C.APrivateAuthKey) -> IO (NonEmpty (QueueId, Either SMPClientError ()))
+smpSubscribeQueues :: SMPSubParty -> SMPClientAgent -> SMPClient -> SMPServer -> NonEmpty (QueueId, C.APrivateAuthKey) -> IO ()
 smpSubscribeQueues party ca smp srv subs = do
-  rs <- L.zip subs <$> subscribe smp (L.map swap subs)
-  atomically $ forM rs $ \(sub, r) ->
-    (fst sub,) <$> case r of
-      Right () -> do
-        addSubscription ca srv $ first (party,) sub
-        pure $ Right ()
-      Left e -> do
-        when (e /= PCENetworkError && e /= PCEResponseTimeout) $
-          removePendingSubscription ca srv (party, fst sub)
-        pure $ Left e
+  rs <- subscribe smp $ L.map swap subs
+  rs' <-
+    atomically $
+      ifM
+        (activeClientSession ca smp srv)
+        (Just <$> processSubscriptions rs)
+        (pure Nothing)
+  case rs' of
+    Just (tempErrs, finalErrs, oks, _) -> do
+      notify_ CASubscribed $ map fst oks
+      notify_ CASubError finalErrs
+      when tempErrs $ reconnectClient ca srv
+    Nothing -> reconnectClient ca srv
   where
+    processSubscriptions :: NonEmpty (Either SMPClientError ()) -> STM (Bool, [(QueueId, SMPClientError)], [(QueueId, (SessionId, C.APrivateAuthKey))], [QueueId])
+    processSubscriptions rs = do
+      pending <- maybe (pure M.empty) readTVar =<< TM.lookup srv (pendingSrvSubs ca)
+      let acc@(_, _, oks, notPending) = foldr (groupSub pending) (False, [], [], []) (L.zip subs rs)
+      unless (null oks) $ addSubscriptions ca srv party oks
+      unless (null notPending) $ removePendingSubs ca srv party notPending
+      pure acc
+    sessId = sessionId $ thParams smp
+    groupSub :: Map SMPSub C.APrivateAuthKey -> ((QueueId, C.APrivateAuthKey), Either SMPClientError ()) -> (Bool, [(QueueId, SMPClientError)], [(QueueId, (SessionId, C.APrivateAuthKey))], [QueueId]) -> (Bool, [(QueueId, SMPClientError)], [(QueueId, (SessionId, C.APrivateAuthKey))], [QueueId])
+    groupSub pending ((qId, pk), r) acc@(!tempErrs, finalErrs, oks, notPending) = case r of
+      Right ()
+        | M.member (party, qId) pending -> (tempErrs, finalErrs, (qId, (sessId, pk)) : oks, qId : notPending)
+        | otherwise -> acc
+      Left e
+        | temporaryClientError e -> (True, finalErrs, oks, notPending)
+        | otherwise -> (tempErrs, (qId, e) : finalErrs, oks, qId : notPending)
     subscribe = case party of
       SPRecipient -> subscribeSMPQueues
       SPNotifier -> subscribeSMPQueuesNtfs
+    notify_ :: (SMPServer -> SMPSubParty -> NonEmpty a -> SMPClientAgentEvent) -> [a] -> IO ()
+    notify_ evt qs = mapM_ (notify ca . evt srv party) $ L.nonEmpty qs
+
+activeClientSession' :: SMPClientAgent -> SessionId -> SMPServer -> STM Bool
+activeClientSession' ca sessId srv = sameSess <$> tryReadSessVar srv (smpClients ca)
+  where
+    sameSess = \case
+      Just (Right (_, smp')) -> sessId == sessionId (thParams smp')
+      _ -> False
+
+activeClientSession :: SMPClientAgent -> SMPClient -> SMPServer -> STM Bool
+activeClientSession ca = activeClientSession' ca . sessionId . thParams
 
 showServer :: SMPServer -> ByteString
 showServer ProtocolServer {host, port} =
   strEncode host <> B.pack (if null port then "" else ':' : port)
 
-smpSubscribe :: SMPClient -> (SMPSub, C.APrivateAuthKey) -> ExceptT SMPClientError IO ()
-smpSubscribe smp ((party, queueId), privKey) = subscribe_ smp privKey queueId
+addSubscriptions :: SMPClientAgent -> SMPServer -> SMPSubParty -> [(QueueId, (SessionId, C.APrivateAuthKey))] -> STM ()
+addSubscriptions = addSubsList_ . srvSubs
+{-# INLINE addSubscriptions #-}
+
+addPendingSubs :: SMPClientAgent -> SMPServer -> SMPSubParty -> [(QueueId, C.APrivateAuthKey)] -> STM ()
+addPendingSubs = addSubsList_ . pendingSrvSubs
+{-# INLINE addPendingSubs #-}
+
+addSubsList_ :: TMap SMPServer (TMap SMPSub s) -> SMPServer -> SMPSubParty -> [(QueueId, s)] -> STM ()
+addSubsList_ subs srv party ss = addSubs_ subs srv ss'
   where
-    subscribe_ = case party of
-      SPRecipient -> subscribeSMPQueue
-      SPNotifier -> subscribeSMPQueueNotifications
+    ss' = M.fromList $ map (first (party,)) ss
 
-addSubscription :: SMPClientAgent -> SMPServer -> (SMPSub, C.APrivateAuthKey) -> STM ()
-addSubscription ca srv sub = do
-  addSub_ (srvSubs ca) srv sub
-  removePendingSubscription ca srv $ fst sub
-
-addPendingSubscription :: SMPClientAgent -> SMPServer -> (SMPSub, C.APrivateAuthKey) -> STM ()
-addPendingSubscription = addSub_ . pendingSrvSubs
-
-addSub_ :: TMap SMPServer (TMap SMPSub C.APrivateAuthKey) -> SMPServer -> (SMPSub, C.APrivateAuthKey) -> STM ()
-addSub_ subs srv (s, key) =
+addSubs_ :: TMap SMPServer (TMap SMPSub s) -> SMPServer -> Map SMPSub s -> STM ()
+addSubs_ subs srv ss =
   TM.lookup srv subs >>= \case
-    Just m -> TM.insert s key m
-    _ -> TM.singleton s key >>= \v -> TM.insert srv v subs
+    Just m -> TM.union ss m
+    _ -> newTVar ss >>= \v -> TM.insert srv v subs
 
 removeSubscription :: SMPClientAgent -> SMPServer -> SMPSub -> STM ()
 removeSubscription = removeSub_ . srvSubs
+{-# INLINE removeSubscription #-}
 
-removePendingSubscription :: SMPClientAgent -> SMPServer -> SMPSub -> STM ()
-removePendingSubscription = removeSub_ . pendingSrvSubs
-
-removeSub_ :: TMap SMPServer (TMap SMPSub C.APrivateAuthKey) -> SMPServer -> SMPSub -> STM ()
+removeSub_ :: TMap SMPServer (TMap SMPSub s) -> SMPServer -> SMPSub -> STM ()
 removeSub_ subs srv s = TM.lookup srv subs >>= mapM_ (TM.delete s)
 
-getSubKey :: TMap SMPServer (TMap SMPSub C.APrivateAuthKey) -> SMPServer -> SMPSub -> STM (Maybe C.APrivateAuthKey)
-getSubKey subs srv s = TM.lookup srv subs $>>= TM.lookup s
+removePendingSubs :: SMPClientAgent -> SMPServer -> SMPSubParty -> [QueueId] -> STM ()
+removePendingSubs = removeSubs_ . pendingSrvSubs
+{-# INLINE removePendingSubs #-}
 
-hasSub :: TMap SMPServer (TMap SMPSub C.APrivateAuthKey) -> SMPServer -> SMPSub -> STM Bool
-hasSub subs srv s = maybe (pure False) (TM.member s) =<< TM.lookup srv subs
+removeSubs_ :: TMap SMPServer (TMap SMPSub C.APrivateAuthKey) -> SMPServer -> SMPSubParty -> [QueueId] -> STM ()
+removeSubs_ subs srv party qs = TM.lookup srv subs >>= mapM_ (`modifyTVar'` (`M.withoutKeys` ss))
+  where
+    ss = S.fromList $ map (party,) qs

src/Simplex/Messaging/Crypto.hs

@@ -152,6 +152,12 @@ module Simplex.Messaging.Crypto
     unsafeSbKey,
     randomSbKey,
 
+    -- * secret_box chains
+    SbChainKey,
+    SbKeyNonce,
+    sbcInit,
+    sbcHkdf,
+
     -- * pseudo-random bytes
     randomBytes,
 
@@ -198,6 +204,7 @@ import qualified Crypto.Cipher.Types as AES
 import qualified Crypto.Cipher.XSalsa as XSalsa
 import qualified Crypto.Error as CE
 import Crypto.Hash (Digest, SHA256 (..), SHA512 (..), hash, hashDigestSize)
+import qualified Crypto.KDF.HKDF as H
 import qualified Crypto.MAC.Poly1305 as Poly1305
 import qualified Crypto.PubKey.Curve25519 as X25519
 import qualified Crypto.PubKey.Curve448 as X448
@@ -1334,6 +1341,26 @@ unsafeSbKey s = either error id $ sbKey s
 randomSbKey :: TVar ChaChaDRG -> STM SbKey
 randomSbKey gVar = SecretBoxKey <$> randomBytes 32 gVar
 
+newtype SbChainKey = SecretBoxChainKey {unSbChainKey :: ByteString}
+  deriving (Eq, Show)
+
+sbcInit :: ByteArrayAccess secret => ByteString -> secret -> (SbChainKey, SbChainKey)
+sbcInit salt secret = (SecretBoxChainKey ck1, SecretBoxChainKey ck2)
+  where
+    prk = H.extract salt secret :: H.PRK SHA512
+    out = H.expand prk ("SimpleXSbChainInit" :: ByteString) 64
+    (ck1, ck2) = B.splitAt 32 out
+
+type SbKeyNonce = (SbKey, CbNonce)
+
+sbcHkdf :: SbChainKey -> (SbKeyNonce, SbChainKey)
+sbcHkdf (SecretBoxChainKey ck) = ((SecretBoxKey sk, CryptoBoxNonce nonce), SecretBoxChainKey ck')
+  where
+    prk = H.extract B.empty ck :: H.PRK SHA512
+    out = H.expand prk ("SimpleXSbChain" :: ByteString) 88 -- = 32 (new chain key) + 32 (secret_box key) + 24 (nonce)
+    (ck', rest) = B.splitAt 32 out
+    (sk, nonce) = B.splitAt 32 rest
+
 xSalsa20 :: ByteArrayAccess key => key -> ByteString -> ByteString -> (ByteString, ByteString)
 xSalsa20 secret nonce msg = (rs, msg')
   where

src/Simplex/Messaging/Crypto/Lazy.hs

@@ -17,6 +17,8 @@ module Simplex.Messaging.Crypto.Lazy
     kcbEncryptTailTag,
     sbDecryptTailTag,
     kcbDecryptTailTag,
+    sbEncryptTailTagNoPad,
+    sbDecryptTailTagNoPad,
     fastReplicate,
     secretBox,
     secretBoxTailTag,
@@ -49,7 +51,7 @@ import Data.Composition ((.:.))
 import Data.Int (Int64)
 import Data.List.NonEmpty (NonEmpty (..))
 import Foreign (sizeOf)
-import Simplex.Messaging.Crypto (CbNonce, CryptoError (..), DhSecret (..), DhSecretX25519, SbKey, pattern CbNonce, pattern SbKey)
+import Simplex.Messaging.Crypto (CbNonce, CryptoError (..), DhSecret (..), DhSecretX25519, SbKey, SbKeyNonce, pattern CbNonce, pattern SbKey)
 import Simplex.Messaging.Crypto.SNTRUP761 (KEMHybridSecret (..))
 import Simplex.Messaging.Encoding
 
@@ -142,6 +144,10 @@ sbEncryptTailTag_ :: ByteArrayAccess key => key -> CbNonce -> LazyByteString ->
 sbEncryptTailTag_ key (CbNonce nonce) msg len paddedLen =
   LB.fromChunks <$> (secretBoxTailTag sbEncryptChunk key nonce =<< pad msg len paddedLen)
 
+sbEncryptTailTagNoPad :: SbKeyNonce -> LazyByteString -> Either CryptoError LazyByteString
+sbEncryptTailTagNoPad (SbKey key, CbNonce nonce) msg =
+  LB.fromChunks <$> secretBoxTailTag sbEncryptChunk key nonce msg
+
 -- | NaCl @secret_box@ decrypt with a symmetric 256-bit key and 192-bit nonce with appended auth tag (more efficient with large files).
 -- paddedLen should NOT include the tag length, it should be the same number that is passed to sbEncrypt / sbEncryptTailTag.
 sbDecryptTailTag :: SbKey -> CbNonce -> Int64 -> LazyByteString -> Either CryptoError (Bool, LazyByteString)
@@ -165,6 +171,15 @@ sbDecryptTailTag_ key (CbNonce nonce) paddedLen packet =
   where
     (c, tag') = LB.splitAt paddedLen packet
 
+sbDecryptTailTagNoPad :: SbKeyNonce -> Int64 -> LazyByteString -> Either CryptoError (Bool, LazyByteString)
+sbDecryptTailTagNoPad (SbKey key, CbNonce nonce) paddedLen packet =
+  result <$> secretBox sbDecryptChunk key nonce c
+  where
+    result (tag :| cs) =
+      let valid = LB.length tag' == 16 && BA.constEq (LB.toStrict tag') tag
+       in (valid, LB.fromChunks cs)
+    (c, tag') = LB.splitAt paddedLen packet
+
 secretBoxTailTag :: ByteArrayAccess key => (SbState -> ByteString -> (ByteString, SbState)) -> key -> ByteString -> LazyByteString -> Either CryptoError [ByteString]
 secretBoxTailTag sbProcess secret nonce msg = run <$> sbInit_ secret nonce
   where

src/Simplex/Messaging/Crypto/SNTRUP761.hs

@@ -4,7 +4,7 @@
 
 module Simplex.Messaging.Crypto.SNTRUP761 where
 
-import Crypto.Hash (Digest, SHA256, hash)
+import Crypto.Hash (Digest, SHA3_256, hash)
 import Data.ByteArray (ScrubbedBytes)
 import qualified Data.ByteArray as BA
 import Data.ByteString (ByteString)
@@ -28,4 +28,4 @@ kcbEncrypt (KEMHybridSecret k) = sbEncrypt_ k
 kemHybridSecret :: PublicKeyX25519 -> PrivateKeyX25519 -> KEMSharedKey -> KEMHybridSecret
 kemHybridSecret k pk (KEMSharedKey kem) =
   let DhSecretX25519 dh = C.dh' k pk
-   in KEMHybridSecret $ BA.convert (hash $ BA.convert dh <> kem :: Digest SHA256)
+   in KEMHybridSecret $ BA.convert (hash $ BA.convert dh <> kem :: Digest SHA3_256)

src/Simplex/Messaging/Encoding.hs

@@ -42,10 +42,12 @@ class Encoding a where
   -- | decoding of type (default implementation uses parser)
   smpDecode :: ByteString -> Either String a
   smpDecode = parseAll smpP
+  {-# INLINE smpDecode #-}
 
   -- | protocol parser of type (default implementation parses protocol ByteString encoding)
   smpP :: Parser a
   smpP = smpDecode <$?> smpP
+  {-# INLINE smpP #-}
 
 instance Encoding Char where
   smpEncode = B.singleton

src/Simplex/Messaging/Encoding/String.hs

@@ -28,6 +28,8 @@ import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Char8 as B
 import Data.Char (isAlphaNum)
 import Data.Int (Int64)
+import Data.IntSet (IntSet)
+import qualified Data.IntSet as IS
 import qualified Data.List.NonEmpty as L
 import Data.Set (Set)
 import qualified Data.Set as S
@@ -39,7 +41,7 @@ import Data.Time.Format.ISO8601
 import Data.Word (Word16, Word32)
 import Simplex.Messaging.Encoding
 import Simplex.Messaging.Parsers (parseAll)
-import Simplex.Messaging.Util ((<$?>))
+import Simplex.Messaging.Util (bshow, (<$?>))
 
 class TextEncoding a where
   textEncode :: a -> Text
@@ -53,14 +55,20 @@ class StrEncoding a where
   -- Please note - if you only specify strDecode, it will use base64urlP as default parser before decoding the string
   strDecode :: ByteString -> Either String a
   strDecode = parseAll strP
+  {-# INLINE strDecode #-}
+
   strP :: Parser a
   strP = strDecode <$?> base64urlP
+  {-# INLINE strP #-}
 
 -- base64url encoding/decoding of ByteStrings - the parser only allows non-empty strings
 instance StrEncoding ByteString where
   strEncode = U.encode
+  {-# INLINE strEncode #-}
   strDecode = U.decode
+  {-# INLINE strDecode #-}
   strP = base64urlP
+  {-# INLINE strP #-}
 
 base64urlP :: Parser ByteString
 base64urlP = do
@@ -119,15 +127,15 @@ instance StrEncoding Bool where
   {-# INLINE strP #-}
 
 instance StrEncoding Int where
-  strEncode = B.pack . show
+  strEncode = bshow
   {-# INLINE strEncode #-}
-  strP = A.decimal
+  strP = A.signed A.decimal
   {-# INLINE strP #-}
 
 instance StrEncoding Int64 where
-  strEncode = B.pack . show
+  strEncode = bshow
   {-# INLINE strEncode #-}
-  strP = A.decimal
+  strP = A.signed A.decimal
   {-# INLINE strP #-}
 
 instance StrEncoding SystemTime where
@@ -154,6 +162,10 @@ instance (StrEncoding a, Ord a) => StrEncoding (Set a) where
   strEncode = strEncodeList . S.toList
   strP = S.fromList <$> listItem `A.sepBy'` A.char ','
 
+instance StrEncoding IntSet where
+  strEncode = strEncodeList . IS.toList
+  strP = IS.fromList <$> listItem `A.sepBy'` A.char ','
+
 listItem :: StrEncoding a => Parser a
 listItem = parseAll strP <$?> A.takeTill (\c -> c == ',' || c == ' ' || c == '\n')
 

src/Simplex/Messaging/Notifications/Client.hs

@@ -1,28 +1,36 @@
 {-# LANGUAGE DataKinds #-}
 {-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE PatternSynonyms #-}
+{-# LANGUAGE TypeApplications #-}
 
 module Simplex.Messaging.Notifications.Client where
 
 import Control.Monad.Except
 import Control.Monad.Trans.Except
+import Data.List.NonEmpty (NonEmpty (..))
+import qualified Data.List.NonEmpty as L
 import Data.Word (Word16)
 import Simplex.Messaging.Client
 import qualified Simplex.Messaging.Crypto as C
 import Simplex.Messaging.Notifications.Protocol
 import Simplex.Messaging.Notifications.Transport (NTFVersion, supportedClientNTFVRange, supportedNTFHandshakes)
-import Simplex.Messaging.Protocol (ErrorType)
+import Simplex.Messaging.Protocol (ErrorType, pattern NoEntity)
+import Simplex.Messaging.Transport (TLS, Transport (..))
 
 type NtfClient = ProtocolClient NTFVersion ErrorType NtfResponse
 
 type NtfClientError = ProtocolClientError ErrorType
 
 defaultNTFClientConfig :: ProtocolClientConfig NTFVersion
-defaultNTFClientConfig = defaultClientConfig (Just supportedNTFHandshakes) supportedClientNTFVRange
+defaultNTFClientConfig =
+  (defaultClientConfig (Just supportedNTFHandshakes) False supportedClientNTFVRange)
+    {defaultTransport = ("443", transport @TLS)}
+{-# INLINE defaultNTFClientConfig #-}
 
 ntfRegisterToken :: NtfClient -> C.APrivateAuthKey -> NewNtfEntity 'Token -> ExceptT NtfClientError IO (NtfTokenId, C.PublicKeyX25519)
 ntfRegisterToken c pKey newTkn =
-  sendNtfCommand c (Just pKey) "" (TNEW newTkn) >>= \case
+  sendNtfCommand c (Just pKey) NoEntity (TNEW newTkn) >>= \case
     NRTknId tknId dhKey -> pure (tknId, dhKey)
     r -> throwE $ unexpectedResponse r
 
@@ -46,16 +54,34 @@ ntfEnableCron c pKey tknId int = okNtfCommand (TCRN int) c pKey tknId
 
 ntfCreateSubscription :: NtfClient -> C.APrivateAuthKey -> NewNtfEntity 'Subscription -> ExceptT NtfClientError IO NtfSubscriptionId
 ntfCreateSubscription c pKey newSub =
-  sendNtfCommand c (Just pKey) "" (SNEW newSub) >>= \case
+  sendNtfCommand c (Just pKey) NoEntity (SNEW newSub) >>= \case
     NRSubId subId -> pure subId
     r -> throwE $ unexpectedResponse r
 
+ntfCreateSubscriptions :: NtfClient -> C.APrivateAuthKey -> NonEmpty (NewNtfEntity 'Subscription) -> IO (NonEmpty (Either NtfClientError NtfSubscriptionId))
+ntfCreateSubscriptions c pKey newSubs = L.map process <$> sendProtocolCommands c cs
+  where
+    cs = L.map (\newSub -> (Just pKey, NoEntity, NtfCmd SSubscription $ SNEW newSub)) newSubs
+    process (Response _ r) = case r of
+      Right (NRSubId subId) -> Right subId
+      Right r' -> Left $ unexpectedResponse r'
+      Left e -> Left e
+
 ntfCheckSubscription :: NtfClient -> C.APrivateAuthKey -> NtfSubscriptionId -> ExceptT NtfClientError IO NtfSubStatus
 ntfCheckSubscription c pKey subId =
   sendNtfCommand c (Just pKey) subId SCHK >>= \case
     NRSub stat -> pure stat
     r -> throwE $ unexpectedResponse r
 
+ntfCheckSubscriptions :: NtfClient -> C.APrivateAuthKey -> NonEmpty NtfSubscriptionId -> IO (NonEmpty (Either NtfClientError NtfSubStatus))
+ntfCheckSubscriptions c pKey subIds = L.map process <$> sendProtocolCommands c cs
+  where
+    cs = L.map (\subId -> (Just pKey, subId, NtfCmd SSubscription SCHK)) subIds
+    process (Response _ r) = case r of
+      Right (NRSub stat) -> Right stat
+      Right r' -> Left $ unexpectedResponse r'
+      Left e -> Left e
+
 ntfDeleteSubscription :: NtfClient -> C.APrivateAuthKey -> NtfSubscriptionId -> ExceptT NtfClientError IO ()
 ntfDeleteSubscription = okNtfCommand SDEL
 

src/Simplex/Messaging/Notifications/Protocol.hs

@@ -20,8 +20,11 @@ import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Char8 as B
 import Data.Functor (($>))
 import Data.Kind
+import Data.List.NonEmpty (NonEmpty (..))
+import qualified Data.List.NonEmpty as L
 import Data.Maybe (isNothing)
 import Data.Text.Encoding (decodeLatin1, encodeUtf8)
+import Data.Time.Clock.System
 import Data.Type.Equality
 import Data.Word (Word16)
 import Database.SQLite.Simple.FromField (FromField (..))
@@ -208,7 +211,7 @@ instance NtfEntityI e => ProtocolEncoding NTFVersion ErrorType (NtfCommand e) wh
   fromProtocolError = fromProtocolError @NTFVersion @ErrorType @NtfResponse
   {-# INLINE fromProtocolError #-}
 
-  checkCredentials (auth, _, entityId, _) cmd = case cmd of
+  checkCredentials (auth, _, EntityId entityId, _) cmd = case cmd of
     -- TNEW and SNEW must have signature but NOT token/subscription IDs
     TNEW {} -> sigNoEntity
     SNEW {} -> sigNoEntity
@@ -322,7 +325,7 @@ instance ProtocolEncoding NTFVersion ErrorType NtfResponse where
     PEBlock -> BLOCK
   {-# INLINE fromProtocolError #-}
 
-  checkCredentials (_, _, entId, _) cmd = case cmd of
+  checkCredentials (_, _, EntityId entId, _) cmd = case cmd of
     -- IDTKN response must not have queue ID
     NRTknId {} -> noEntity
     -- IDSUB response must not have queue ID
@@ -426,7 +429,31 @@ instance FromJSON DeviceToken where
     t <- encodeUtf8 <$> o .: "token"
     pure $ DeviceToken pp t
 
-type NtfEntityId = ByteString
+-- List of PNMessageData uses semicolon-separated encoding instead of strEncode,
+-- because strEncode of NonEmpty list uses comma for separator,
+-- and encoding of PNMessageData's smpQueue has comma in list of hosts
+encodePNMessages :: NonEmpty PNMessageData -> ByteString
+encodePNMessages = B.intercalate ";" . map strEncode . L.toList
+
+pnMessagesP :: A.Parser (NonEmpty PNMessageData)
+pnMessagesP = L.fromList <$> strP `A.sepBy1` A.char ';'
+
+data PNMessageData = PNMessageData
+  { smpQueue :: SMPQueueNtf,
+    ntfTs :: SystemTime,
+    nmsgNonce :: C.CbNonce,
+    encNMsgMeta :: EncNMsgMeta
+  }
+  deriving (Show)
+
+instance StrEncoding PNMessageData where
+  strEncode PNMessageData {smpQueue, ntfTs, nmsgNonce, encNMsgMeta} =
+    strEncode (smpQueue, ntfTs, nmsgNonce, encNMsgMeta)
+  strP = do
+    (smpQueue, ntfTs, nmsgNonce, encNMsgMeta) <- strP
+    pure PNMessageData {smpQueue, ntfTs, nmsgNonce, encNMsgMeta}
+
+type NtfEntityId = EntityId
 
 type NtfSubscriptionId = NtfEntityId
 
@@ -443,6 +470,8 @@ data NtfSubStatus
     NSInactive
   | -- | END received
     NSEnd
+  | -- | DELD received (connection was deleted)
+    NSDeleted
   | -- | SMP AUTH error
     NSAuth
   | -- | SMP error other than AUTH
@@ -456,6 +485,7 @@ ntfShouldSubscribe = \case
   NSActive -> True
   NSInactive -> True
   NSEnd -> False
+  NSDeleted -> False
   NSAuth -> False
   NSErr _ -> False
 
@@ -466,6 +496,7 @@ instance Encoding NtfSubStatus where
     NSActive -> "ACTIVE"
     NSInactive -> "INACTIVE"
     NSEnd -> "END"
+    NSDeleted -> "DELETED"
     NSAuth -> "AUTH"
     NSErr err -> "ERR " <> err
   smpP =
@@ -475,6 +506,7 @@ instance Encoding NtfSubStatus where
       "ACTIVE" -> pure NSActive
       "INACTIVE" -> pure NSInactive
       "END" -> pure NSEnd
+      "DELETED" -> pure NSDeleted
       "AUTH" -> pure NSAuth
       "ERR" -> NSErr <$> (A.space *> A.takeByteString)
       _ -> fail "bad NtfSubStatus"

src/Simplex/Messaging/Notifications/Server.hs

@@ -1,3 +1,5 @@
+{-# LANGUAGE CPP #-}
+{-# LANGUAGE BangPatterns #-}
 {-# LANGUAGE DataKinds #-}
 {-# LANGUAGE DuplicateRecordFields #-}
 {-# LANGUAGE FlexibleContexts #-}
@@ -7,6 +9,7 @@
 {-# LANGUAGE NumericUnderscores #-}
 {-# LANGUAGE OverloadedLists #-}
 {-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE PatternSynonyms #-}
 {-# LANGUAGE ScopedTypeVariables #-}
 {-# LANGUAGE TupleSections #-}
 
@@ -16,11 +19,17 @@ import Control.Logger.Simple
 import Control.Monad
 import Control.Monad.Except
 import Control.Monad.Reader
+import Data.Bifunctor (first)
+import qualified Data.ByteString.Builder as BLD
 import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Char8 as B
+import qualified Data.ByteString.Lazy.Char8 as LB
+import Data.Either (partitionEithers)
 import Data.Functor (($>))
+import Data.IORef
 import Data.Int (Int64)
-import Data.List (intercalate, sort)
+import qualified Data.IntSet as IS
+import Data.List (intercalate, partition, sort)
 import Data.List.NonEmpty (NonEmpty (..))
 import qualified Data.List.NonEmpty as L
 import qualified Data.Map.Strict as M
@@ -30,34 +39,43 @@ import Data.Text.Encoding (decodeLatin1)
 import Data.Time.Clock (UTCTime (..), diffTimeToPicoseconds, getCurrentTime)
 import Data.Time.Clock.System (getSystemTime)
 import Data.Time.Format.ISO8601 (iso8601Show)
-import Network.Socket (ServiceName)
+import GHC.IORef (atomicSwapIORef)
+import GHC.Stats (getRTSStats)
+import Network.Socket (ServiceName, Socket, socketToHandle)
 import Simplex.Messaging.Client (ProtocolClientError (..), SMPClientError, ServerTransmission (..))
 import Simplex.Messaging.Client.Agent
 import qualified Simplex.Messaging.Crypto as C
 import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Notifications.Protocol
+import Simplex.Messaging.Notifications.Server.Control
 import Simplex.Messaging.Notifications.Server.Env
-import Simplex.Messaging.Notifications.Server.Push.APNS (PNMessageData (..), PushNotification (..), PushProviderError (..))
+import Simplex.Messaging.Notifications.Server.Push.APNS (PushNotification (..), PushProviderError (..))
 import Simplex.Messaging.Notifications.Server.Stats
 import Simplex.Messaging.Notifications.Server.Store
 import Simplex.Messaging.Notifications.Server.StoreLog
 import Simplex.Messaging.Notifications.Transport
-import Simplex.Messaging.Protocol (ErrorType (..), ProtocolServer (host), SMPServer, SignedTransmission, Transmission, encodeTransmission, tGet, tPut)
+import Simplex.Messaging.Protocol (EntityId (..), ErrorType (..), ProtocolServer (host), SMPServer, SignedTransmission, Transmission, pattern NoEntity, pattern SMPServer, encodeTransmission, tGet, tPut)
 import qualified Simplex.Messaging.Protocol as SMP
 import Simplex.Messaging.Server
-import Simplex.Messaging.Server.Stats
+import Simplex.Messaging.Server.Control (CPClientRole (..))
+import Simplex.Messaging.Server.Stats (PeriodStats (..), PeriodStatCounts (..), periodStatCounts, updatePeriodStats)
+import Simplex.Messaging.TMap (TMap)
 import qualified Simplex.Messaging.TMap as TM
-import Simplex.Messaging.Transport (ATransport (..), THandle (..), THandleAuth (..), THandleParams (..), TProxy, Transport (..), TransportPeer (..))
-import Simplex.Messaging.Transport.Server (runTransportServer, tlsServerCredentials)
+import Simplex.Messaging.Transport (ATransport (..), THandle (..), THandleAuth (..), THandleParams (..), TProxy, Transport (..), TransportPeer (..), defaultSupportedParams)
+import Simplex.Messaging.Transport.Buffer (trimCR)
+import Simplex.Messaging.Transport.Server (AddHTTP, runTransportServer, runLocalTCPServer)
 import Simplex.Messaging.Util
 import System.Exit (exitFailure)
-import System.IO (BufferMode (..), hPutStrLn, hSetBuffering)
+import System.IO (BufferMode (..), hClose, hPrint, hPutStrLn, hSetBuffering, hSetNewlineMode, universalNewlineMode)
 import System.Mem.Weak (deRefWeak)
-import UnliftIO (IOMode (..), async, uninterruptibleCancel, withFile)
+import UnliftIO (IOMode (..), UnliftIO, askUnliftIO, async, uninterruptibleCancel, unliftIO, withFile)
 import UnliftIO.Concurrent (forkIO, killThread, mkWeakThreadId)
 import UnliftIO.Directory (doesFileExist, renameFile)
 import UnliftIO.Exception
 import UnliftIO.STM
+#if MIN_VERSION_base(4,18,0)
+import GHC.Conc (listThreads)
+#endif
 
 runNtfServer :: NtfServerConfig -> IO ()
 runNtfServer cfg = do
@@ -71,18 +89,19 @@ type M a = ReaderT NtfEnv IO a
 
 ntfServer :: NtfServerConfig -> TMVar Bool -> M ()
 ntfServer cfg@NtfServerConfig {transports, transportConfig = tCfg} started = do
+  restoreServerLastNtfs
   restoreServerStats
   s <- asks subscriber
   ps <- asks pushServer
   resubscribe s
-  raceAny_ (ntfSubscriber s : ntfPush ps : map runServer transports <> serverStatsThread_ cfg) `finally` stopServer
+  raceAny_ (ntfSubscriber s : ntfPush ps : map runServer transports <> serverStatsThread_ cfg <> controlPortThread_ cfg) `finally` stopServer
   where
-    runServer :: (ServiceName, ATransport) -> M ()
-    runServer (tcpPort, ATransport t) = do
-      serverParams <- asks tlsServerParams
-      serverSignKey <- either fail pure . fromTLSCredentials $ tlsServerCredentials serverParams
+    runServer :: (ServiceName, ATransport, AddHTTP) -> M ()
+    runServer (tcpPort, ATransport t, _addHTTP) = do
+      srvCreds <- asks tlsServerCreds
+      serverSignKey <- either fail pure $ fromTLSCredentials srvCreds
       env <- ask
-      liftIO $ runTransportServer started tcpPort serverParams tCfg $ \h -> runClient serverSignKey t h `runReaderT` env
+      liftIO $ runTransportServer started tcpPort defaultSupportedParams srvCreds (Just supportedNTFHandshakes) tCfg $ \h -> runClient serverSignKey t h `runReaderT` env
     fromTLSCredentials (_, pk) = C.x509ToPrivate (pk, []) >>= C.privKey
 
     runClient :: Transport c => C.APrivateSignKey -> TProxy c -> c -> M ()
@@ -96,11 +115,15 @@ ntfServer cfg@NtfServerConfig {transports, transportConfig = tCfg} started = do
 
     stopServer :: M ()
     stopServer = do
-      withNtfLog closeStoreLog
-      saveServerStats
+      logInfo "Saving server state..."
+      saveServer
       NtfSubscriber {smpSubscribers, smpAgent} <- asks subscriber
       liftIO $ readTVarIO smpSubscribers >>= mapM_ (\SMPSubscriber {subThreadId} -> readTVarIO subThreadId >>= mapM_ (deRefWeak >=> mapM_ killThread))
       liftIO $ closeSMPClientAgent smpAgent
+      logInfo "Server stopped"
+
+    saveServer :: M ()
+    saveServer = withNtfLog closeStoreLog >> saveServerLastNtfs >> saveServerStats
 
     serverStatsThread_ :: NtfServerConfig -> [M ()]
     serverStatsThread_ NtfServerConfig {logStatsInterval = Just interval, logStatsStartTime, serverStatsLogFile} =
@@ -118,16 +141,16 @@ ntfServer cfg@NtfServerConfig {transports, transportConfig = tCfg} started = do
         withFile statsFilePath AppendMode $ \h -> liftIO $ do
           hSetBuffering h LineBuffering
           ts <- getCurrentTime
-          fromTime' <- atomically $ swapTVar fromTime ts
-          tknCreated' <- atomically $ swapTVar tknCreated 0
-          tknVerified' <- atomically $ swapTVar tknVerified 0
-          tknDeleted' <- atomically $ swapTVar tknDeleted 0
-          subCreated' <- atomically $ swapTVar subCreated 0
-          subDeleted' <- atomically $ swapTVar subDeleted 0
-          ntfReceived' <- atomically $ swapTVar ntfReceived 0
-          ntfDelivered' <- atomically $ swapTVar ntfDelivered 0
-          tkn <- atomically $ periodStatCounts activeTokens ts
-          sub <- atomically $ periodStatCounts activeSubs ts
+          fromTime' <- atomicSwapIORef fromTime ts
+          tknCreated' <- atomicSwapIORef tknCreated 0
+          tknVerified' <- atomicSwapIORef tknVerified 0
+          tknDeleted' <- atomicSwapIORef tknDeleted 0
+          subCreated' <- atomicSwapIORef subCreated 0
+          subDeleted' <- atomicSwapIORef subDeleted 0
+          ntfReceived' <- atomicSwapIORef ntfReceived 0
+          ntfDelivered' <- atomicSwapIORef ntfDelivered 0
+          tkn <- liftIO $ periodStatCounts activeTokens ts
+          sub <- liftIO $ periodStatCounts activeSubs ts
           hPutStrLn h $
             intercalate
               ","
@@ -148,6 +171,131 @@ ntfServer cfg@NtfServerConfig {transports, transportConfig = tCfg} started = do
               ]
         liftIO $ threadDelay' interval
 
+    controlPortThread_ :: NtfServerConfig -> [M ()]
+    controlPortThread_ NtfServerConfig {controlPort = Just port} = [runCPServer port]
+    controlPortThread_ _ = []
+
+    runCPServer :: ServiceName -> M ()
+    runCPServer port = do
+      cpStarted <- newEmptyTMVarIO
+      u <- askUnliftIO
+      liftIO $ do
+        labelMyThread "control port server"
+        runLocalTCPServer cpStarted port $ runCPClient u
+      where
+        runCPClient :: UnliftIO (ReaderT NtfEnv IO) -> Socket -> IO ()
+        runCPClient u sock = do
+          labelMyThread "control port client"
+          h <- socketToHandle sock ReadWriteMode
+          hSetBuffering h LineBuffering
+          hSetNewlineMode h universalNewlineMode
+          hPutStrLn h "Ntf server control port\n'help' for supported commands"
+          role <- newTVarIO CPRNone
+          cpLoop h role
+          where
+            cpLoop h role = do
+              s <- trimCR <$> B.hGetLine h
+              case strDecode s of
+                Right CPQuit -> hClose h
+                Right cmd -> logCmd s cmd >> processCP h role cmd >> cpLoop h role
+                Left err -> hPutStrLn h ("error: " <> err) >> cpLoop h role
+            logCmd s cmd = when shouldLog $ logWarn $ "ControlPort: " <> tshow s
+              where
+                shouldLog = case cmd of
+                  CPAuth _ -> False
+                  CPHelp -> False
+                  CPQuit -> False
+                  CPSkip -> False
+                  _ -> True
+            processCP h role = \case
+              CPAuth auth -> atomically $ writeTVar role $! newRole cfg
+                where
+                  newRole NtfServerConfig {controlPortUserAuth = user, controlPortAdminAuth = admin}
+                    | Just auth == admin = CPRAdmin
+                    | Just auth == user = CPRUser
+                    | otherwise = CPRNone
+              CPStats -> withUserRole $ do
+                ss <- unliftIO u $ asks serverStats
+                let getStat :: (NtfServerStats -> IORef a) -> IO a
+                    getStat var = readIORef (var ss)
+                    putStat :: Show a => String -> (NtfServerStats -> IORef a) -> IO ()
+                    putStat label var = getStat var >>= \v -> hPutStrLn h $ label <> ": " <> show v
+                putStat "fromTime" fromTime
+                putStat "tknCreated" tknCreated
+                putStat "tknVerified" tknVerified
+                putStat "tknDeleted" tknDeleted
+                putStat "subCreated" subCreated
+                putStat "subDeleted" subDeleted
+                putStat "ntfReceived" ntfReceived
+                getStat (day . activeTokens) >>= \v -> hPutStrLn h $ "daily active tokens: " <> show (IS.size v)
+                getStat (day . activeSubs) >>= \v -> hPutStrLn h $ "daily active subscriptions: " <> show (IS.size v)
+              CPStatsRTS -> tryAny getRTSStats >>= either (hPrint h) (hPrint h)
+              CPServerInfo -> readTVarIO role >>= \case
+                CPRNone -> do
+                  logError "Unauthorized control port command"
+                  hPutStrLn h "AUTH"
+                r -> do
+#if MIN_VERSION_base(4,18,0)
+                  threads <- liftIO listThreads
+                  hPutStrLn h $ "Threads: " <> show (length threads)
+#else
+                  hPutStrLn h "Threads: not available on GHC 8.10"
+#endif
+                  NtfSubscriber {smpSubscribers, smpAgent = a} <- unliftIO u $ asks subscriber
+                  putSMPWorkers a "SMP subcscribers" smpSubscribers
+                  let SMPClientAgent {smpClients, smpSessions, srvSubs, pendingSrvSubs, smpSubWorkers} = a
+                  putSMPWorkers a "SMP clients" smpClients
+                  putSMPWorkers a "SMP subscription workers" smpSubWorkers
+                  sessions <- readTVarIO smpSessions
+                  hPutStrLn h $ "SMP sessions count: " <> show (M.size sessions)
+                  putSMPSubs a "SMP subscriptions" srvSubs
+                  putSMPSubs a "Pending SMP subscriptions" pendingSrvSubs
+                  where
+                    putSMPSubs :: SMPClientAgent -> String -> TMap SMPServer (TMap SMPSub a) -> IO ()
+                    putSMPSubs a name v = do
+                      subs <- readTVarIO v
+                      (totalCnt, ownCount, otherCnt, servers, ownByServer) <- foldM countSubs (0, 0, 0, [], M.empty) $ M.assocs subs
+                      showServers a name servers
+                      hPutStrLn h $ name <> " total: " <> show totalCnt
+                      hPutStrLn h $ name <> " on own servers: " <> show ownCount
+                      when (r == CPRAdmin && not (null ownByServer)) $
+                        forM_ (M.assocs ownByServer) $ \(SMPServer (host :| _) _ _, cnt) ->
+                          hPutStrLn h $ name <> " on " <> B.unpack (strEncode host) <> ": " <> show cnt
+                      hPutStrLn h $ name <> " on other servers: " <> show otherCnt
+                      where
+                        countSubs :: (Int, Int, Int, [SMPServer], M.Map SMPServer Int) -> (SMPServer, TMap SMPSub a) -> IO (Int, Int, Int, [SMPServer], M.Map SMPServer Int)
+                        countSubs (!totalCnt, !ownCount, !otherCnt, !servers, !ownByServer) (srv, srvSubs) = do
+                          cnt <- M.size <$> readTVarIO srvSubs
+                          let totalCnt' = totalCnt + cnt
+                              ownServer = isOwnServer a srv
+                              (ownCount', otherCnt')
+                                | ownServer = (ownCount + cnt, otherCnt)
+                                | otherwise = (ownCount, otherCnt + cnt)
+                              servers' = if cnt > 0 then srv : servers else servers
+                              ownByServer'
+                                | r == CPRAdmin && ownServer && cnt > 0 = M.alter (Just . maybe cnt (+ cnt)) srv ownByServer
+                                | otherwise = ownByServer
+                          pure (totalCnt', ownCount', otherCnt', servers', ownByServer')
+                    putSMPWorkers :: SMPClientAgent -> String -> TMap SMPServer a -> IO ()
+                    putSMPWorkers a name v = readTVarIO v >>= showServers a name . M.keys
+                    showServers :: SMPClientAgent -> String -> [SMPServer] -> IO ()
+                    showServers a name srvs = do
+                      let (ownSrvs, otherSrvs) = partition (isOwnServer a) srvs
+                      hPutStrLn h $ name <> " own servers count: " <> show (length ownSrvs)
+                      when (r == CPRAdmin) $ hPutStrLn h $ name <> " own servers: " <> intercalate "," (sort $ map (\(SMPServer (host :| _) _ _) -> B.unpack $ strEncode host) ownSrvs)
+                      hPutStrLn h $ name <> " other servers count: " <> show (length otherSrvs)
+              CPHelp -> hPutStrLn h "commands: stats, stats-rts, server-info, help, quit"
+              CPQuit -> pure ()
+              CPSkip -> pure ()
+              where
+                withUserRole action =
+                  readTVarIO role >>= \case
+                    CPRAdmin -> action
+                    CPRUser -> action
+                    _ -> do
+                      logError "Unauthorized control port command"
+                      hPutStrLn h "AUTH"
+
 resubscribe :: NtfSubscriber -> M ()
 resubscribe NtfSubscriber {newSubQ} = do
   logInfo "Preparing SMP resubscriptions..."
@@ -176,10 +324,10 @@ ntfSubscriber NtfSubscriber {smpSubscribers, newSubQ, smpAgent = ca@SMPClientAge
 
     getSMPSubscriber :: SMPServer -> M SMPSubscriber
     getSMPSubscriber smpServer =
-      atomically (TM.lookup smpServer smpSubscribers) >>= maybe createSMPSubscriber pure
+      liftIO (TM.lookupIO smpServer smpSubscribers) >>= maybe createSMPSubscriber pure
       where
         createSMPSubscriber = do
-          sub@SMPSubscriber {subThreadId} <- atomically newSMPSubscriber
+          sub@SMPSubscriber {subThreadId} <- liftIO newSMPSubscriber
           atomically $ TM.insert smpServer sub smpSubscribers
           tId <- mkWeakThreadId =<< forkIO (runSMPSubscriber sub)
           atomically . writeTVar subThreadId $ Just tId
@@ -188,39 +336,22 @@ ntfSubscriber NtfSubscriber {smpSubscribers, newSubQ, smpAgent = ca@SMPClientAge
     runSMPSubscriber :: SMPSubscriber -> M ()
     runSMPSubscriber SMPSubscriber {newSubQ = subscriberSubQ} =
       forever $ do
-        subs <- atomically (peekTQueue subscriberSubQ)
+        subs <- atomically $ readTQueue subscriberSubQ
         let subs' = L.map (\(NtfSub sub) -> sub) subs
             srv = server $ L.head subs
         logSubStatus srv "subscribing" $ length subs
         mapM_ (\NtfSubData {smpQueue} -> updateSubStatus smpQueue NSPending) subs'
-        rs <- liftIO $ subscribeQueues srv subs'
-        (subs'', oks, errs) <- foldM process ([], 0, []) rs
-        atomically $ do
-          void $ readTQueue subscriberSubQ
-          mapM_ (writeTQueue subscriberSubQ . L.map NtfSub) $ L.nonEmpty subs''
-        logSubStatus srv "retrying" $ length subs''
-        logSubStatus srv "subscribed" oks
-        logSubErrors srv errs
-      where
-        process :: ([NtfSubData], Int, [NtfSubStatus]) -> (NtfSubData, Either SMPClientError ()) -> M ([NtfSubData], Int, [NtfSubStatus])
-        process (subs, oks, errs) (sub@NtfSubData {smpQueue}, r) = case r of
-          Right _ -> updateSubStatus smpQueue NSActive $> (subs, oks + 1, errs)
-          Left e -> update <$> handleSubError smpQueue e
-          where
-            update = \case
-              Just err -> (subs, oks, err : errs) -- permanent error, log and don't retry subscription
-              Nothing -> (sub : subs, oks, errs) -- temporary error, retry subscription
+        liftIO $ subscribeQueues srv subs'
 
     -- \| Subscribe to queues. The list of results can have a different order.
-    subscribeQueues :: SMPServer -> NonEmpty NtfSubData -> IO (NonEmpty (NtfSubData, Either SMPClientError ()))
-    subscribeQueues srv subs =
-      L.zipWith (\s r -> (s, snd r)) subs <$> subscribeQueuesNtfs ca srv (L.map sub subs)
+    subscribeQueues :: SMPServer -> NonEmpty NtfSubData -> IO ()
+    subscribeQueues srv subs = subscribeQueuesNtfs ca srv (L.map sub subs)
       where
         sub NtfSubData {smpQueue = SMPQueueNtf {notifierId}, notifierKey} = (notifierId, notifierKey)
 
     receiveSMP :: M ()
     receiveSMP = forever $ do
-      ((_, srv, _), _, _, ts) <- atomically $ readTBQueue msgQ
+      ((_, srv, _), _thVersion, sessionId, ts) <- atomically $ readTBQueue msgQ
       forM ts $ \(ntfId, t) -> case t of
         STUnexpectedError e -> logError $ "SMP client unexpected error: " <> tshow e -- uncorrelated response, should not happen
         STResponse {} -> pure () -- it was already reported as timeout error
@@ -232,14 +363,19 @@ ntfSubscriber NtfSubscriber {smpSubscribers, newSubQ, smpAgent = ca@SMPClientAge
               st <- asks store
               NtfPushServer {pushQ} <- asks pushServer
               stats <- asks serverStats
-              atomically $ updatePeriodStats (activeSubs stats) ntfId
-              atomically $
-                findNtfSubscriptionToken st smpQueue
-                  >>= mapM_ (\tkn -> writeTBQueue pushQ (tkn, PNMessage PNMessageData {smpQueue, ntfTs, nmsgNonce, encNMsgMeta}))
+              liftIO $ updatePeriodStats (activeSubs stats) ntfId
+              tkn_ <- atomically (findNtfSubscriptionToken st smpQueue)
+              forM_ tkn_ $ \tkn -> do
+                let newNtf = PNMessageData {smpQueue, ntfTs, nmsgNonce, encNMsgMeta}
+                lastNtfs <- liftIO $ addTokenLastNtf st (ntfTknId tkn) newNtf
+                atomically (writeTBQueue pushQ (tkn, PNMessage lastNtfs))
               incNtfStat ntfReceived
-            Right SMP.END -> updateSubStatus smpQueue NSEnd
+            Right SMP.END ->
+              whenM (atomically $ activeClientSession' ca sessionId srv) $
+                updateSubStatus smpQueue NSEnd
+            Right SMP.DELD -> updateSubStatus smpQueue NSDeleted
             Right (SMP.ERR e) -> logError $ "SMP server error: " <> tshow e
-            Right _ -> logError $ "SMP server unexpected response"
+            Right _ -> logError "SMP server unexpected response"
             Left e -> logError $ "SMP client error: " <> tshow e
 
     receiveAgent =
@@ -252,11 +388,11 @@ ntfSubscriber NtfSubscriber {smpSubscribers, newSubQ, smpAgent = ca@SMPClientAge
             forM_ subs $ \(_, ntfId) -> do
               let smpQueue = SMPQueueNtf srv ntfId
               updateSubStatus smpQueue NSInactive
-          CAResubscribed srv subs -> do
-            forM_ subs $ \(_, ntfId) -> updateSubStatus (SMPQueueNtf srv ntfId) NSActive
-            logSubStatus srv "resubscribed" $ length subs
-          CASubError srv errs ->
-            forM errs (\((_, ntfId), err) -> handleSubError (SMPQueueNtf srv ntfId) err)
+          CASubscribed srv _ subs -> do
+            forM_ subs $ \ntfId -> updateSubStatus (SMPQueueNtf srv ntfId) NSActive
+            logSubStatus srv "subscribed" $ length subs
+          CASubError srv _ errs ->
+            forM errs (\(ntfId, err) -> handleSubError (SMPQueueNtf srv ntfId) err)
               >>= logSubErrors srv . catMaybes . L.toList
 
     logSubStatus srv event n =
@@ -314,7 +450,7 @@ ntfPush s@NtfPushServer {pushQ} = forever $ do
       void $ deliverNotification pp tkn ntf
     PNMessage {} -> checkActiveTkn status $ do
       stats <- asks serverStats
-      atomically $ updatePeriodStats (activeTokens stats) ntfTknId
+      liftIO $ updatePeriodStats (activeTokens stats) ntfTknId
       void $ deliverNotification pp tkn ntf
       incNtfStat ntfDelivered
   where
@@ -350,7 +486,7 @@ runNtfClientTransport :: Transport c => THandleNTF c 'TServer -> M ()
 runNtfClientTransport th@THandle {params} = do
   qSize <- asks $ clientQSize . config
   ts <- liftIO getSystemTime
-  c <- atomically $ newNtfServerClient qSize params ts
+  c <- liftIO $ newNtfServerClient qSize params ts
   s <- asks subscriber
   ps <- asks pushServer
   expCfg <- asks $ inactiveClientExpiration . config
@@ -365,28 +501,29 @@ clientDisconnected NtfServerClient {connected} = atomically $ writeTVar connecte
 
 receive :: Transport c => THandleNTF c 'TServer -> NtfServerClient -> M ()
 receive th@THandle {params = THandleParams {thAuth}} NtfServerClient {rcvQ, sndQ, rcvActiveAt} = forever $ do
-  ts <- liftIO $ tGet th
-  forM_ ts $ \t@(_, _, (corrId, entId, cmdOrError)) -> do
-    atomically . writeTVar rcvActiveAt =<< liftIO getSystemTime
-    logDebug "received transmission"
-    case cmdOrError of
-      Left e -> write sndQ (corrId, entId, NRErr e)
-      Right cmd ->
-        verifyNtfTransmission ((,C.cbNonce (SMP.bs corrId)) <$> thAuth) t cmd >>= \case
-          VRVerified req -> write rcvQ req
-          VRFailed -> write sndQ (corrId, entId, NRErr AUTH)
+  ts <- L.toList <$> liftIO (tGet th)
+  atomically . (writeTVar rcvActiveAt $!) =<< liftIO getSystemTime
+  (errs, cmds) <- partitionEithers <$> mapM cmdAction ts
+  write sndQ errs
+  write rcvQ cmds
   where
-    write q t = atomically $ writeTBQueue q t
+    cmdAction t@(_, _, (corrId, entId, cmdOrError)) =
+      case cmdOrError of
+        Left e -> pure $ Left (corrId, entId, NRErr e)
+        Right cmd ->
+          verified <$> verifyNtfTransmission ((,C.cbNonce (SMP.bs corrId)) <$> thAuth) t cmd
+          where
+            verified = \case
+              VRVerified req -> Right req
+              VRFailed -> Left (corrId, entId, NRErr AUTH)
+    write q = mapM_ (atomically . writeTBQueue q) . L.nonEmpty
 
 send :: Transport c => THandleNTF c 'TServer -> NtfServerClient -> IO ()
 send h@THandle {params} NtfServerClient {sndQ, sndActiveAt} = forever $ do
-  t <- atomically $ readTBQueue sndQ
-  void . liftIO $ tPut h [Right (Nothing, encodeTransmission params t)]
+  ts <- atomically $ readTBQueue sndQ
+  void . liftIO $ tPut h $ L.map (\t -> Right (Nothing, encodeTransmission params t)) ts
   atomically . (writeTVar sndActiveAt $!) =<< liftIO getSystemTime
 
--- instance Show a => Show (TVar a) where
---   show x = unsafePerformIO $ show <$> readTVarIO x
-
 data VerificationResult = VRVerified NtfRequest | VRFailed
 
 verifyNtfTransmission :: Maybe (THandleAuth 'TServer, C.CbNonce) -> SignedTransmission ErrorType NtfCmd -> NtfCmd -> M VerificationResult
@@ -444,7 +581,7 @@ client :: NtfServerClient -> NtfSubscriber -> NtfPushServer -> M ()
 client NtfServerClient {rcvQ, sndQ} NtfSubscriber {newSubQ, smpAgent = ca} NtfPushServer {pushQ, intervalNotifiers} =
   forever $
     atomically (readTBQueue rcvQ)
-      >>= processCommand
+      >>= mapM processCommand
       >>= atomically . writeTBQueue sndQ
   where
     processCommand :: NtfRequest -> M (Transmission NtfResponse)
@@ -461,7 +598,7 @@ client NtfServerClient {rcvQ, sndQ} NtfSubscriber {newSubQ, smpAgent = ca} NtfPu
         atomically $ writeTBQueue pushQ (tkn, PNVerification regCode)
         withNtfLog (`logCreateToken` tkn)
         incNtfStatT token tknCreated
-        pure (corrId, "", NRTknId tknId srvDhPubKey)
+        pure (corrId, NoEntity, NRTknId tknId srvDhPubKey)
       NtfReqCmd SToken (NtfTkn tkn@NtfTknData {token, ntfTknId, tknStatus, tknRegCode, tknDhSecret, tknDhKeys = (srvDhPubKey, srvDhPrivKey), tknCronInterval}) (corrId, tknId, cmd) -> do
         status <- readTVarIO tknStatus
         (corrId,tknId,) <$> case cmd of
@@ -524,7 +661,7 @@ client NtfServerClient {rcvQ, sndQ} NtfSubscriber {newSubQ, smpAgent = ca} NtfPu
             | otherwise -> do
                 logDebug "TCRN"
                 atomically $ writeTVar tknCronInterval int
-                atomically (TM.lookup tknId intervalNotifiers) >>= \case
+                liftIO (TM.lookupIO tknId intervalNotifiers) >>= \case
                   Nothing -> runIntervalNotifier int
                   Just IntervalNotifier {interval, action} ->
                     unless (interval == int) $ do
@@ -552,7 +689,7 @@ client NtfServerClient {rcvQ, sndQ} NtfSubscriber {newSubQ, smpAgent = ca} NtfPu
             _ -> pure $ NRErr AUTH
         withNtfLog (`logCreateSubscription` sub)
         incNtfStat subCreated
-        pure (corrId, "", resp)
+        pure (corrId, NoEntity, resp)
       NtfReqCmd SSubscription (NtfSub NtfSubData {smpQueue = SMPQueueNtf {smpServer, notifierId}, notifierKey = registeredNKey, subStatus}) (corrId, subId, cmd) -> do
         status <- readTVarIO subStatus
         (corrId,subId,) <$> case cmd of
@@ -577,7 +714,7 @@ client NtfServerClient {rcvQ, sndQ} NtfSubscriber {newSubQ, smpAgent = ca} NtfPu
           PING -> pure NRPong
       NtfReqPing corrId entId -> pure (corrId, entId, NRPong)
     getId :: M NtfEntityId
-    getId = randomBytes =<< asks (subIdBytes . config)
+    getId = fmap EntityId . randomBytes =<< asks (subIdBytes . config)
     getRegCode :: M NtfRegCode
     getRegCode = NtfRegCode <$> (randomBytes =<< asks (regCodeBytes . config))
     randomBytes :: Int -> M ByteString
@@ -590,19 +727,57 @@ client NtfServerClient {rcvQ, sndQ} NtfSubscriber {newSubQ, smpAgent = ca} NtfPu
 withNtfLog :: (StoreLog 'WriteMode -> IO a) -> M ()
 withNtfLog action = liftIO . mapM_ action =<< asks storeLog
 
-incNtfStatT :: DeviceToken -> (NtfServerStats -> TVar Int) -> M ()
+incNtfStatT :: DeviceToken -> (NtfServerStats -> IORef Int) -> M ()
 incNtfStatT (DeviceToken PPApnsNull _) _ = pure ()
 incNtfStatT _ statSel = incNtfStat statSel
 
-incNtfStat :: (NtfServerStats -> TVar Int) -> M ()
+incNtfStat :: (NtfServerStats -> IORef Int) -> M ()
 incNtfStat statSel = do
   stats <- asks serverStats
-  atomically $ modifyTVar' (statSel stats) (+ 1)
+  liftIO $ atomicModifyIORef'_ (statSel stats) (+ 1)
+
+saveServerLastNtfs :: M ()
+saveServerLastNtfs = asks (storeLastNtfsFile . config) >>= mapM_ saveLastNtfs
+  where
+    saveLastNtfs f = do
+      logInfo $ "saving last notifications to file " <> T.pack f
+      NtfStore {tokenLastNtfs} <- asks store
+      liftIO . withFile f WriteMode $ \h ->
+        readTVarIO tokenLastNtfs >>= mapM_ (saveTokenLastNtfs h) . M.assocs
+      logInfo "notifications saved"
+      where
+        -- reverse on save, to save notifications in order, will become reversed again when restoring.
+        saveTokenLastNtfs h (tknId, v) = BLD.hPutBuilder h . encodeLastNtfs tknId . L.reverse =<< readTVarIO v
+        encodeLastNtfs tknId = mconcat . L.toList . L.map (\ntf -> BLD.byteString (strEncode $ TNMRv1 tknId ntf) <> BLD.char8 '\n')
+
+restoreServerLastNtfs :: M ()
+restoreServerLastNtfs =
+  asks (storeLastNtfsFile . config) >>= mapM_ restoreLastNtfs
+  where
+    restoreLastNtfs f =
+      whenM (doesFileExist f) $ do
+        logInfo $ "restoring last notifications from file " <> T.pack f
+        st <- asks store
+        runExceptT (liftIO (LB.readFile f) >>= mapM (restoreNtf st) . LB.lines) >>= \case
+          Left e -> do
+            logError . T.pack $ "error restoring last notifications: " <> e
+            liftIO exitFailure
+          Right _ -> do
+            renameFile f $ f <> ".bak"
+            logInfo "last notifications restored"
+      where
+        restoreNtf st s' = do
+          TNMRv1 tknId ntf <- liftEither . first (ntfErr "parsing") $ strDecode s
+          liftIO $ storeTokenLastNtf st tknId ntf
+          where
+            s = LB.toStrict s'
+            ntfErr :: Show e => String -> e -> String
+            ntfErr op e = op <> " error (" <> show e <> "): " <> B.unpack (B.take 100 s)
 
 saveServerStats :: M ()
 saveServerStats =
   asks (serverStatsBackupFile . config)
-    >>= mapM_ (\f -> asks serverStats >>= atomically . getNtfServerStatsData >>= liftIO . saveStats f)
+    >>= mapM_ (\f -> asks serverStats >>= liftIO . getNtfServerStatsData >>= liftIO . saveStats f)
   where
     saveStats f stats = do
       logInfo $ "saving server stats to file " <> T.pack f
@@ -617,7 +792,7 @@ restoreServerStats = asks (serverStatsBackupFile . config) >>= mapM_ restoreStat
       liftIO (strDecode <$> B.readFile f) >>= \case
         Right d -> do
           s <- asks serverStats
-          atomically $ setNtfServerStats s d
+          liftIO $ setNtfServerStats s d
           renameFile f $ f <> ".bak"
           logInfo "server stats restored"
         Left e -> do

src/Simplex/Messaging/Notifications/Server/Control.hs

@@ -0,0 +1,37 @@
+{-# LANGUAGE LambdaCase #-}
+{-# LANGUAGE OverloadedStrings #-}
+
+module Simplex.Messaging.Notifications.Server.Control where
+
+import qualified Data.Attoparsec.ByteString.Char8 as A
+import Simplex.Messaging.Encoding.String
+import Simplex.Messaging.Protocol (BasicAuth)
+
+data ControlProtocol
+  = CPAuth BasicAuth
+  | CPStats
+  | CPStatsRTS
+  | CPServerInfo
+  | CPHelp
+  | CPQuit
+  | CPSkip
+
+instance StrEncoding ControlProtocol where
+  strEncode = \case
+    CPAuth tok -> "auth " <> strEncode tok
+    CPStats -> "stats"
+    CPStatsRTS -> "stats-rts"
+    CPServerInfo -> "server-info"
+    CPHelp -> "help"
+    CPQuit -> "quit"
+    CPSkip -> ""
+  strP =
+    A.takeTill (== ' ') >>= \case
+      "auth" -> CPAuth <$> _strP
+      "stats" -> pure CPStats
+      "stats-rts" -> pure CPStatsRTS
+      "server-info" -> pure CPServerInfo
+      "help" -> pure CPHelp
+      "quit" -> pure CPQuit
+      "" -> pure CPSkip
+      _ -> fail "bad ControlProtocol command"

src/Simplex/Messaging/Notifications/Server/Env.hs

@@ -10,7 +10,6 @@ module Simplex.Messaging.Notifications.Server.Env where
 import Control.Concurrent (ThreadId)
 import Control.Concurrent.Async (Async)
 import Control.Logger.Simple
-import Control.Monad.IO.Unlift
 import Crypto.Random
 import Data.Int (Int64)
 import Data.List.NonEmpty (NonEmpty)
@@ -29,18 +28,21 @@ import Simplex.Messaging.Notifications.Server.Stats
 import Simplex.Messaging.Notifications.Server.Store
 import Simplex.Messaging.Notifications.Server.StoreLog
 import Simplex.Messaging.Notifications.Transport (NTFVersion, VersionRangeNTF)
-import Simplex.Messaging.Protocol (CorrId, SMPServer, Transmission)
+import Simplex.Messaging.Protocol (BasicAuth, CorrId, SMPServer, Transmission)
 import Simplex.Messaging.Server.Expiration
 import Simplex.Messaging.TMap (TMap)
 import qualified Simplex.Messaging.TMap as TM
 import Simplex.Messaging.Transport (ATransport, THandleParams, TransportPeer (..))
-import Simplex.Messaging.Transport.Server (TransportServerConfig, alpn, loadFingerprint, loadTLSServerParams)
+import Simplex.Messaging.Transport.Server (AddHTTP, ServerCredentials, TransportServerConfig, loadFingerprint, loadServerCredential)
 import System.IO (IOMode (..))
 import System.Mem.Weak (Weak)
 import UnliftIO.STM
 
 data NtfServerConfig = NtfServerConfig
-  { transports :: [(ServiceName, ATransport)],
+  { transports :: [(ServiceName, ATransport, AddHTTP)],
+    controlPort :: Maybe ServiceName,
+    controlPortUserAuth :: Maybe BasicAuth,
+    controlPortAdminAuth :: Maybe BasicAuth,
     subIdBytes :: Int,
     regCodeBytes :: Int,
     clientQSize :: Natural,
@@ -51,10 +53,8 @@ data NtfServerConfig = NtfServerConfig
     subsBatchSize :: Int,
     inactiveClientExpiration :: Maybe ExpirationConfig,
     storeLogFile :: Maybe FilePath,
-    -- CA certificate private key is not needed for initialization
-    caCertificateFile :: FilePath,
-    privateKeyFile :: FilePath,
-    certificateFile :: FilePath,
+    storeLastNtfsFile :: Maybe FilePath,
+    ntfCredentials :: ServerCredentials,
     -- stats config - see SMP server config
     logStatsInterval :: Maybe Int64,
     logStatsStartTime :: Int64,
@@ -78,24 +78,24 @@ data NtfEnv = NtfEnv
     store :: NtfStore,
     storeLog :: Maybe (StoreLog 'WriteMode),
     random :: TVar ChaChaDRG,
-    tlsServerParams :: T.ServerParams,
+    tlsServerCreds :: T.Credential,
     serverIdentity :: C.KeyHash,
     serverStats :: NtfServerStats
   }
 
 newNtfServerEnv :: NtfServerConfig -> IO NtfEnv
-newNtfServerEnv config@NtfServerConfig {subQSize, pushQSize, smpAgentCfg, apnsConfig, storeLogFile, caCertificateFile, certificateFile, privateKeyFile, transportConfig} = do
-  random <- liftIO C.newRandom
-  store <- atomically newNtfStore
+newNtfServerEnv config@NtfServerConfig {subQSize, pushQSize, smpAgentCfg, apnsConfig, storeLogFile, ntfCredentials} = do
+  random <- C.newRandom
+  store <- newNtfStore
   logInfo "restoring subscriptions..."
-  storeLog <- liftIO $ mapM (`readWriteNtfStore` store) storeLogFile
+  storeLog <- mapM (`readWriteNtfStore` store) storeLogFile
   logInfo "restored subscriptions"
-  subscriber <- atomically $ newNtfSubscriber subQSize smpAgentCfg random
-  pushServer <- atomically $ newNtfPushServer pushQSize apnsConfig
-  tlsServerParams <- liftIO $ loadTLSServerParams caCertificateFile certificateFile privateKeyFile (alpn transportConfig)
-  Fingerprint fp <- liftIO $ loadFingerprint caCertificateFile
-  serverStats <- atomically . newNtfServerStats =<< liftIO getCurrentTime
-  pure NtfEnv {config, subscriber, pushServer, store, storeLog, random, tlsServerParams, serverIdentity = C.KeyHash fp, serverStats}
+  subscriber <- newNtfSubscriber subQSize smpAgentCfg random
+  pushServer <- newNtfPushServer pushQSize apnsConfig
+  tlsServerCreds <- loadServerCredential ntfCredentials
+  Fingerprint fp <- loadFingerprint ntfCredentials
+  serverStats <- newNtfServerStats =<< getCurrentTime
+  pure NtfEnv {config, subscriber, pushServer, store, storeLog, random, tlsServerCreds, serverIdentity = C.KeyHash fp, serverStats}
 
 data NtfSubscriber = NtfSubscriber
   { smpSubscribers :: TMap SMPServer SMPSubscriber,
@@ -103,10 +103,10 @@ data NtfSubscriber = NtfSubscriber
     smpAgent :: SMPClientAgent
   }
 
-newNtfSubscriber :: Natural -> SMPClientAgentConfig -> TVar ChaChaDRG -> STM NtfSubscriber
+newNtfSubscriber :: Natural -> SMPClientAgentConfig -> TVar ChaChaDRG -> IO NtfSubscriber
 newNtfSubscriber qSize smpAgentCfg random = do
-  smpSubscribers <- TM.empty
-  newSubQ <- newTBQueue qSize
+  smpSubscribers <- TM.emptyIO
+  newSubQ <- newTBQueueIO qSize
   smpAgent <- newSMPClientAgent smpAgentCfg random
   pure NtfSubscriber {smpSubscribers, newSubQ, smpAgent}
 
@@ -115,10 +115,10 @@ data SMPSubscriber = SMPSubscriber
     subThreadId :: TVar (Maybe (Weak ThreadId))
   }
 
-newSMPSubscriber :: STM SMPSubscriber
+newSMPSubscriber :: IO SMPSubscriber
 newSMPSubscriber = do
-  newSubQ <- newTQueue
-  subThreadId <- newTVar Nothing
+  newSubQ <- newTQueueIO
+  subThreadId <- newTVarIO Nothing
   pure SMPSubscriber {newSubQ, subThreadId}
 
 data NtfPushServer = NtfPushServer
@@ -134,11 +134,11 @@ data IntervalNotifier = IntervalNotifier
     interval :: Word16
   }
 
-newNtfPushServer :: Natural -> APNSPushClientConfig -> STM NtfPushServer
+newNtfPushServer :: Natural -> APNSPushClientConfig -> IO NtfPushServer
 newNtfPushServer qSize apnsConfig = do
-  pushQ <- newTBQueue qSize
-  pushClients <- TM.empty
-  intervalNotifiers <- TM.empty
+  pushQ <- newTBQueueIO qSize
+  pushClients <- TM.emptyIO
+  intervalNotifiers <- TM.emptyIO
   pure NtfPushServer {pushQ, pushClients, intervalNotifiers, apnsConfig}
 
 newPushClient :: NtfPushServer -> PushProvider -> IO PushProviderClient
@@ -151,7 +151,7 @@ newPushClient NtfPushServer {apnsConfig, pushClients} pp = do
 
 getPushClient :: NtfPushServer -> PushProvider -> IO PushProviderClient
 getPushClient s@NtfPushServer {pushClients} pp =
-  atomically (TM.lookup pp pushClients) >>= maybe (newPushClient s pp) pure
+  TM.lookupIO pp pushClients >>= maybe (newPushClient s pp) pure
 
 data NtfRequest
   = NtfReqNew CorrId ANewNtfEntity
@@ -159,19 +159,19 @@ data NtfRequest
   | NtfReqPing CorrId NtfEntityId
 
 data NtfServerClient = NtfServerClient
-  { rcvQ :: TBQueue NtfRequest,
-    sndQ :: TBQueue (Transmission NtfResponse),
+  { rcvQ :: TBQueue (NonEmpty NtfRequest),
+    sndQ :: TBQueue (NonEmpty (Transmission NtfResponse)),
     ntfThParams :: THandleParams NTFVersion 'TServer,
     connected :: TVar Bool,
     rcvActiveAt :: TVar SystemTime,
     sndActiveAt :: TVar SystemTime
   }
 
-newNtfServerClient :: Natural -> THandleParams NTFVersion 'TServer -> SystemTime -> STM NtfServerClient
+newNtfServerClient :: Natural -> THandleParams NTFVersion 'TServer -> SystemTime -> IO NtfServerClient
 newNtfServerClient qSize ntfThParams ts = do
-  rcvQ <- newTBQueue qSize
-  sndQ <- newTBQueue qSize
-  connected <- newTVar True
-  rcvActiveAt <- newTVar ts
-  sndActiveAt <- newTVar ts
+  rcvQ <- newTBQueueIO qSize
+  sndQ <- newTBQueueIO qSize
+  connected <- newTVarIO True
+  rcvActiveAt <- newTVarIO ts
+  sndActiveAt <- newTVarIO ts
   return NtfServerClient {rcvQ, sndQ, ntfThParams, connected, rcvActiveAt, sndActiveAt}

src/Simplex/Messaging/Notifications/Server/Main.hs

@@ -1,31 +1,36 @@
 {-# LANGUAGE DuplicateRecordFields #-}
 {-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE NumericUnderscores #-}
 {-# LANGUAGE OverloadedLists #-}
 {-# LANGUAGE OverloadedStrings #-}
 {-# LANGUAGE PatternSynonyms #-}
+{-# OPTIONS_GHC -fno-warn-ambiguous-fields #-}
 
 module Simplex.Messaging.Notifications.Server.Main where
 
+import Control.Monad ((<$!>))
 import Data.Functor (($>))
 import Data.Ini (lookupValue, readIniFile)
 import Data.Maybe (fromMaybe)
 import qualified Data.Text as T
+import Data.Text.Encoding (encodeUtf8)
 import qualified Data.Text.IO as T
 import Network.Socket (HostName)
 import Options.Applicative
+import Simplex.Messaging.Client (HostMode (..), NetworkConfig (..), ProtocolClientConfig (..), SocksMode (..), defaultNetworkConfig, textToHostMode)
 import Simplex.Messaging.Client.Agent (SMPClientAgentConfig (..), defaultSMPClientAgentConfig)
 import qualified Simplex.Messaging.Crypto as C
 import Simplex.Messaging.Notifications.Server (runNtfServer)
 import Simplex.Messaging.Notifications.Server.Env (NtfServerConfig (..), defaultInactiveClientExpiration)
 import Simplex.Messaging.Notifications.Server.Push.APNS (defaultAPNSPushClientConfig)
-import Simplex.Messaging.Notifications.Transport (supportedNTFHandshakes, supportedServerNTFVRange)
+import Simplex.Messaging.Notifications.Transport (supportedServerNTFVRange)
 import Simplex.Messaging.Protocol (ProtoServerWithAuth (..), pattern NtfServer)
 import Simplex.Messaging.Server.CLI
 import Simplex.Messaging.Server.Expiration
 import Simplex.Messaging.Transport (simplexMQVersion)
 import Simplex.Messaging.Transport.Client (TransportHost (..))
-import Simplex.Messaging.Transport.Server (TransportServerConfig (..), defaultTransportServerConfig)
+import Simplex.Messaging.Transport.Server (ServerCredentials (..), TransportServerConfig (..), defaultTransportServerConfig)
 import Simplex.Messaging.Util (tshow)
 import System.Directory (createDirectoryIfMissing, doesFileExist)
 import System.FilePath (combine)
@@ -48,7 +53,9 @@ ntfServerCLI cfgPath logPath =
         True -> readIniFile iniFile >>= either exitError runServer
         _ -> exitError $ "Error: server is not initialized (" <> iniFile <> " does not exist).\nRun `" <> executableName <> " init`."
     Delete -> do
-      confirmOrExit "WARNING: deleting the server will make all queues inaccessible, because the server identity (certificate fingerprint) will change.\nTHIS CANNOT BE UNDONE!"
+      confirmOrExit
+        "WARNING: deleting the server will make all queues inaccessible, because the server identity (certificate fingerprint) will change.\nTHIS CANNOT BE UNDONE!"
+        "Server NOT deleted"
       deleteDirIfExists cfgPath
       deleteDirIfExists logPath
       putStrLn "Deleted configuration and log files"
@@ -80,13 +87,38 @@ ntfServerCLI cfgPath logPath =
           \# and restoring it when the server is started.\n\
           \# Log is compacted on start (deleted objects are removed).\n"
             <> ("enable: " <> onOff enableStoreLog <> "\n\n")
+            <> "# Last notifications are optionally saved and restored when the server restarts,\n\
+               \# they are preserved in the .bak file until the next restart.\n"
+            <> ("restore_last_notifications: " <> onOff enableStoreLog <> "\n\n")
             <> "log_stats: off\n\n\
+               \[AUTH]\n\
+               \# control_port_admin_password:\n\
+               \# control_port_user_password:\n\
+               \\n\
                \[TRANSPORT]\n\
-               \# host is only used to print server address on start\n"
+               \# Host is only used to print server address on start.\n\
+               \# You can specify multiple server ports.\n"
             <> ("host: " <> T.pack host <> "\n")
             <> ("port: " <> T.pack defaultServerPort <> "\n")
-            <> "log_tls_errors: off\n"
-            <> "websockets: off\n\n\
+            <> "log_tls_errors: off\n\n\
+               \# Use `websockets: 443` to run websockets server in addition to plain TLS.\n\
+               \websockets: off\n\n\
+               \# control_port: 5227\n\
+               \\n\
+               \[SUBSCRIBER]\n\
+               \# Network configuration for notification server client.\n\
+               \# `host_mode` can be 'public' (default) or 'onion'.\n\
+               \# It defines prefferred hostname for destination servers with multiple hostnames.\n\
+               \# host_mode: public\n\
+               \# required_host_mode: off\n\n\
+               \# SOCKS proxy port for subscribing to SMP servers.\n\
+               \# You may need a separate instance of SOCKS proxy for incoming single-hop requests.\n\
+               \# socks_proxy: localhost:9050\n\n\
+               \# `socks_mode` can be 'onion' for SOCKS proxy to be used for .onion destination hosts only (default)\n\
+               \# or 'always' to be used for all destination hosts (can be used if it is an .onion server).\n\
+               \# socks_mode: onion\n\n\
+               \# The domain suffixes of the relays you operate (space-separated) to count as separate proxy statistics.\n\
+               \# own_server_domains: \n\n\
                \[INACTIVE_CLIENTS]\n\
                \# TTL and interval to check inactive clients\n\
                \disconnect: off\n"
@@ -107,15 +139,38 @@ ntfServerCLI cfgPath logPath =
         enableStoreLog = settingIsOn "STORE_LOG" "enable" ini
         logStats = settingIsOn "STORE_LOG" "log_stats" ini
         c = combine cfgPath . ($ defaultX509Config)
+        restoreLastNtfsFile path = case iniOnOff "STORE_LOG" "restore_last_notifications" ini of
+          Just True -> Just path
+          Just False -> Nothing
+          -- if the setting is not set, it is enabled when store log is enabled
+          _ -> enableStoreLog $> path
         serverConfig =
           NtfServerConfig
             { transports = iniTransports ini,
+              controlPort = either (const Nothing) (Just . T.unpack) $ lookupValue "TRANSPORT" "control_port" ini,
+              controlPortAdminAuth = either error id <$!> strDecodeIni "AUTH" "control_port_admin_password" ini,
+              controlPortUserAuth = either error id <$!> strDecodeIni "AUTH" "control_port_user_password" ini,
               subIdBytes = 24,
               regCodeBytes = 32,
               clientQSize = 64,
               subQSize = 512,
               pushQSize = 1048,
-              smpAgentCfg = defaultSMPClientAgentConfig {persistErrorInterval = 0},
+              smpAgentCfg =
+                defaultSMPClientAgentConfig
+                  { smpCfg =
+                      (smpCfg defaultSMPClientAgentConfig)
+                        { networkConfig =
+                            defaultNetworkConfig
+                              { socksProxy = either error id <$!> strDecodeIni "SUBSCRIBER" "socks_proxy" ini,
+                                socksMode = maybe SMOnion (either error id) $! strDecodeIni "SUBSCRIBER" "socks_mode" ini,
+                                hostMode = either (const HMPublic) (either error id . textToHostMode) $ lookupValue "SUBSCRIBER" "host_mode" ini,
+                                requiredHostMode = fromMaybe False $ iniOnOff "SUBSCRIBER" "required_host_mode" ini,
+                                smpPingInterval = 60_000_000 -- 1 minute
+                              }
+                        },
+                    ownServerDomains = either (const []) (map encodeUtf8 . T.words) $ lookupValue "SUBSCRIBER" "own_server_domains" ini,
+                    persistErrorInterval = 0 -- seconds
+                  },
               apnsConfig = defaultAPNSPushClientConfig,
               subsBatchSize = 900,
               inactiveClientExpiration =
@@ -125,9 +180,13 @@ ntfServerCLI cfgPath logPath =
                       checkInterval = readStrictIni "INACTIVE_CLIENTS" "check_interval" ini
                     },
               storeLogFile = enableStoreLog $> storeLogFilePath,
-              caCertificateFile = c caCrtFile,
-              privateKeyFile = c serverKeyFile,
-              certificateFile = c serverCrtFile,
+              storeLastNtfsFile = restoreLastNtfsFile $ combine logPath "ntf-server-last-notifications.log",
+              ntfCredentials =
+                ServerCredentials
+                  { caCertificateFile = Just $ c caCrtFile,
+                    privateKeyFile = c serverKeyFile,
+                    certificateFile = c serverCrtFile
+                  },
               logStatsInterval = logStats $> 86400, -- seconds
               logStatsStartTime = 0, -- seconds from 00:00 UTC
               serverStatsLogFile = combine logPath "ntf-server-stats.daily.log",
@@ -135,8 +194,7 @@ ntfServerCLI cfgPath logPath =
               ntfServerVRange = supportedServerNTFVRange,
               transportConfig =
                 defaultTransportServerConfig
-                  { logTLSErrors = fromMaybe False $ iniOnOff "TRANSPORT" "log_tls_errors" ini,
-                    alpn = Just supportedNTFHandshakes
+                  { logTLSErrors = fromMaybe False $ iniOnOff "TRANSPORT" "log_tls_errors" ini
                   }
             }
 

src/Simplex/Messaging/Notifications/Server/Push/APNS.hs

@@ -34,6 +34,7 @@ import Data.ByteString.Builder (lazyByteString)
 import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Lazy.Char8 as LB
 import Data.Int (Int64)
+import Data.List.NonEmpty (NonEmpty (..))
 import Data.Map.Strict (Map)
 import Data.Maybe (isNothing)
 import Data.Text (Text)
@@ -47,12 +48,10 @@ import Network.HTTP2.Client (Request)
 import qualified Network.HTTP2.Client as H
 import Network.Socket (HostName, ServiceName)
 import qualified Simplex.Messaging.Crypto as C
-import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Notifications.Protocol
 import Simplex.Messaging.Notifications.Server.Push.APNS.Internal
 import Simplex.Messaging.Notifications.Server.Store (NtfTknData (..))
 import Simplex.Messaging.Parsers (defaultJSON)
-import Simplex.Messaging.Protocol (EncNMsgMeta)
 import Simplex.Messaging.Transport.HTTP2 (HTTP2Body (..))
 import Simplex.Messaging.Transport.HTTP2.Client
 import Simplex.Messaging.Util (safeDecodeUtf8, tshow)
@@ -103,26 +102,11 @@ readECPrivateKey f = do
 
 data PushNotification
   = PNVerification NtfRegCode
-  | PNMessage PNMessageData
+  | PNMessage (NonEmpty PNMessageData)
   | -- | PNAlert Text
     PNCheckMessages
   deriving (Show)
 
-data PNMessageData = PNMessageData
-  { smpQueue :: SMPQueueNtf,
-    ntfTs :: SystemTime,
-    nmsgNonce :: C.CbNonce,
-    encNMsgMeta :: EncNMsgMeta
-  }
-  deriving (Show)
-
-instance StrEncoding PNMessageData where
-  strEncode PNMessageData {smpQueue, ntfTs, nmsgNonce, encNMsgMeta} =
-    strEncode (smpQueue, ntfTs, nmsgNonce, encNMsgMeta)
-  strP = do
-    (smpQueue, ntfTs, nmsgNonce, encNMsgMeta) <- strP
-    pure PNMessageData {smpQueue, ntfTs, nmsgNonce, encNMsgMeta}
-
 data APNSNotification = APNSNotification {aps :: APNSNotificationBody, notificationData :: Maybe J.Value}
   deriving (Show)
 
@@ -208,7 +192,7 @@ defaultAPNSPushClientConfig =
       authKeyFileEnv = "APNS_KEY_FILE", -- the environment variables APNS_KEY_FILE and APNS_KEY_ID must be set, or the server would fail to start
       authKeyAlg = "ES256",
       authKeyIdEnv = "APNS_KEY_ID",
-      paddedNtfLength = 512,
+      paddedNtfLength = 3072,
       appName = "chat.simplex.app",
       appTeamId = "5NN7GUYB6T",
       apnsPort = "443",
@@ -285,7 +269,7 @@ apnsNotification NtfTknData {tknDhSecret} nonce paddedLen = \case
     encrypt code $ \code' ->
       apn APNSBackground {contentAvailable = 1} . Just $ J.object ["nonce" .= nonce, "verification" .= code']
   PNMessage pnMessageData ->
-    encrypt (strEncode pnMessageData) $ \ntfData ->
+    encrypt (encodePNMessages pnMessageData) $ \ntfData ->
       apn apnMutableContent . Just $ J.object ["nonce" .= nonce, "message" .= ntfData]
   -- PNAlert text -> Right $ apn (apnAlert $ APNSAlertText text) Nothing
   PNCheckMessages -> Right $ apn APNSBackground {contentAvailable = 1} . Just $ J.object ["checkMessages" .= True]

src/Simplex/Messaging/Notifications/Server/Stats.hs

@@ -7,24 +7,22 @@ module Simplex.Messaging.Notifications.Server.Stats where
 import Control.Applicative (optional)
 import qualified Data.Attoparsec.ByteString.Char8 as A
 import qualified Data.ByteString.Char8 as B
+import Data.IORef
 import Data.Time.Clock (UTCTime)
 import Simplex.Messaging.Encoding.String
-import Simplex.Messaging.Notifications.Protocol (NtfTokenId)
-import Simplex.Messaging.Protocol (NotifierId)
 import Simplex.Messaging.Server.Stats
-import UnliftIO.STM
 
 data NtfServerStats = NtfServerStats
-  { fromTime :: TVar UTCTime,
-    tknCreated :: TVar Int,
-    tknVerified :: TVar Int,
-    tknDeleted :: TVar Int,
-    subCreated :: TVar Int,
-    subDeleted :: TVar Int,
-    ntfReceived :: TVar Int,
-    ntfDelivered :: TVar Int,
-    activeTokens :: PeriodStats NtfTokenId,
-    activeSubs :: PeriodStats NotifierId
+  { fromTime :: IORef UTCTime,
+    tknCreated :: IORef Int,
+    tknVerified :: IORef Int,
+    tknDeleted :: IORef Int,
+    subCreated :: IORef Int,
+    subDeleted :: IORef Int,
+    ntfReceived :: IORef Int,
+    ntfDelivered :: IORef Int,
+    activeTokens :: PeriodStats,
+    activeSubs :: PeriodStats
   }
 
 data NtfServerStatsData = NtfServerStatsData
@@ -36,48 +34,49 @@ data NtfServerStatsData = NtfServerStatsData
     _subDeleted :: Int,
     _ntfReceived :: Int,
     _ntfDelivered :: Int,
-    _activeTokens :: PeriodStatsData NtfTokenId,
-    _activeSubs :: PeriodStatsData NotifierId
+    _activeTokens :: PeriodStatsData,
+    _activeSubs :: PeriodStatsData
   }
 
-newNtfServerStats :: UTCTime -> STM NtfServerStats
+newNtfServerStats :: UTCTime -> IO NtfServerStats
 newNtfServerStats ts = do
-  fromTime <- newTVar ts
-  tknCreated <- newTVar 0
-  tknVerified <- newTVar 0
-  tknDeleted <- newTVar 0
-  subCreated <- newTVar 0
-  subDeleted <- newTVar 0
-  ntfReceived <- newTVar 0
-  ntfDelivered <- newTVar 0
+  fromTime <- newIORef ts
+  tknCreated <- newIORef 0
+  tknVerified <- newIORef 0
+  tknDeleted <- newIORef 0
+  subCreated <- newIORef 0
+  subDeleted <- newIORef 0
+  ntfReceived <- newIORef 0
+  ntfDelivered <- newIORef 0
   activeTokens <- newPeriodStats
   activeSubs <- newPeriodStats
   pure NtfServerStats {fromTime, tknCreated, tknVerified, tknDeleted, subCreated, subDeleted, ntfReceived, ntfDelivered, activeTokens, activeSubs}
 
-getNtfServerStatsData :: NtfServerStats -> STM NtfServerStatsData
+getNtfServerStatsData :: NtfServerStats -> IO NtfServerStatsData
 getNtfServerStatsData s@NtfServerStats {fromTime} = do
-  _fromTime <- readTVar fromTime
-  _tknCreated <- readTVar $ tknCreated s
-  _tknVerified <- readTVar $ tknVerified s
-  _tknDeleted <- readTVar $ tknDeleted s
-  _subCreated <- readTVar $ subCreated s
-  _subDeleted <- readTVar $ subDeleted s
-  _ntfReceived <- readTVar $ ntfReceived s
-  _ntfDelivered <- readTVar $ ntfDelivered s
+  _fromTime <- readIORef fromTime
+  _tknCreated <- readIORef $ tknCreated s
+  _tknVerified <- readIORef $ tknVerified s
+  _tknDeleted <- readIORef $ tknDeleted s
+  _subCreated <- readIORef $ subCreated s
+  _subDeleted <- readIORef $ subDeleted s
+  _ntfReceived <- readIORef $ ntfReceived s
+  _ntfDelivered <- readIORef $ ntfDelivered s
   _activeTokens <- getPeriodStatsData $ activeTokens s
   _activeSubs <- getPeriodStatsData $ activeSubs s
   pure NtfServerStatsData {_fromTime, _tknCreated, _tknVerified, _tknDeleted, _subCreated, _subDeleted, _ntfReceived, _ntfDelivered, _activeTokens, _activeSubs}
 
-setNtfServerStats :: NtfServerStats -> NtfServerStatsData -> STM ()
+-- this function is not thread safe, it is used on server start only
+setNtfServerStats :: NtfServerStats -> NtfServerStatsData -> IO ()
 setNtfServerStats s@NtfServerStats {fromTime} d@NtfServerStatsData {_fromTime} = do
-  writeTVar fromTime $! _fromTime
-  writeTVar (tknCreated s) $! _tknCreated d
-  writeTVar (tknVerified s) $! _tknVerified d
-  writeTVar (tknDeleted s) $! _tknDeleted d
-  writeTVar (subCreated s) $! _subCreated d
-  writeTVar (subDeleted s) $! _subDeleted d
-  writeTVar (ntfReceived s) $! _ntfReceived d
-  writeTVar (ntfDelivered s) $! _ntfDelivered d
+  writeIORef fromTime $! _fromTime
+  writeIORef (tknCreated s) $! _tknCreated d
+  writeIORef (tknVerified s) $! _tknVerified d
+  writeIORef (tknDeleted s) $! _tknDeleted d
+  writeIORef (subCreated s) $! _subCreated d
+  writeIORef (subDeleted s) $! _subDeleted d
+  writeIORef (ntfReceived s) $! _ntfReceived d
+  writeIORef (ntfDelivered s) $! _ntfDelivered d
   setPeriodStats (activeTokens s) (_activeTokens d)
   setPeriodStats (activeSubs s) (_activeSubs d)
 

src/Simplex/Messaging/Notifications/Server/Store.hs

@@ -1,9 +1,12 @@
+{-# LANGUAGE BangPatterns #-}
 {-# LANGUAGE DataKinds #-}
 {-# LANGUAGE DuplicateRecordFields #-}
 {-# LANGUAGE GADTs #-}
 {-# LANGUAGE KindSignatures #-}
 {-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE OverloadedLists #-}
+{-# LANGUAGE OverloadedStrings #-}
 {-# LANGUAGE ScopedTypeVariables #-}
 
 module Simplex.Messaging.Notifications.Server.Store where
@@ -12,12 +15,14 @@ import Control.Concurrent.STM
 import Control.Monad
 import Data.ByteString.Char8 (ByteString)
 import Data.Functor (($>))
+import Data.List.NonEmpty (NonEmpty (..), (<|))
 import qualified Data.Map.Strict as M
 import Data.Maybe (catMaybes)
 import Data.Set (Set)
 import qualified Data.Set as S
 import Data.Word (Word16)
 import qualified Simplex.Messaging.Crypto as C
+import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Notifications.Protocol
 import Simplex.Messaging.Protocol (NtfPrivateAuthKey, NtfPublicAuthKey, SMPServer)
 import Simplex.Messaging.TMap (TMap)
@@ -30,17 +35,19 @@ data NtfStore = NtfStore
     tokenRegistrations :: TMap DeviceToken (TMap ByteString NtfTokenId),
     subscriptions :: TMap NtfSubscriptionId NtfSubData,
     tokenSubscriptions :: TMap NtfTokenId (TVar (Set NtfSubscriptionId)),
-    subscriptionLookup :: TMap SMPQueueNtf NtfSubscriptionId
+    subscriptionLookup :: TMap SMPQueueNtf NtfSubscriptionId,
+    tokenLastNtfs :: TMap NtfTokenId (TVar (NonEmpty PNMessageData))
   }
 
-newNtfStore :: STM NtfStore
+newNtfStore :: IO NtfStore
 newNtfStore = do
-  tokens <- TM.empty
-  tokenRegistrations <- TM.empty
-  subscriptions <- TM.empty
-  tokenSubscriptions <- TM.empty
-  subscriptionLookup <- TM.empty
-  pure NtfStore {tokens, tokenRegistrations, subscriptions, tokenSubscriptions, subscriptionLookup}
+  tokens <- TM.emptyIO
+  tokenRegistrations <- TM.emptyIO
+  subscriptions <- TM.emptyIO
+  tokenSubscriptions <- TM.emptyIO
+  subscriptionLookup <- TM.emptyIO
+  tokenLastNtfs <- TM.emptyIO
+  pure NtfStore {tokens, tokenRegistrations, subscriptions, tokenSubscriptions, subscriptionLookup, tokenLastNtfs}
 
 data NtfTknData = NtfTknData
   { ntfTknId :: NtfTokenId,
@@ -77,6 +84,9 @@ data NtfEntityRec (e :: NtfEntity) where
 getNtfToken :: NtfStore -> NtfTokenId -> STM (Maybe NtfTknData)
 getNtfToken st tknId = TM.lookup tknId (tokens st)
 
+getNtfTokenIO :: NtfStore -> NtfTokenId -> IO (Maybe NtfTknData)
+getNtfTokenIO st tknId = TM.lookupIO tknId (tokens st)
+
 addNtfToken :: NtfStore -> NtfTokenId -> NtfTknData -> STM ()
 addNtfToken st tknId tkn@NtfTknData {token, tknVerifyKey} = do
   TM.insert tknId tkn $ tokens st
@@ -108,6 +118,7 @@ removeInactiveTokenRegistrations st NtfTknData {ntfTknId = tId, token} =
       forM_ tIds $ \(regKey, tId') -> do
         TM.delete regKey tknRegs
         TM.delete tId' $ tokens st
+        TM.delete tId' $ tokenLastNtfs st
         void $ deleteTokenSubs st tId'
       pure $ map snd tIds
 
@@ -127,6 +138,7 @@ deleteNtfToken st tknId = do
       TM.lookup token regs $>>= \tIds ->
         TM.delete (regKey tknVerifyKey) tIds
           >> whenM (TM.null tIds) (TM.delete token regs) $> Just ()
+  TM.delete tknId $ tokenLastNtfs st
   deleteTokenSubs st tknId
   where
     regs = tokenRegistrations st
@@ -193,3 +205,42 @@ deleteNtfSubscription st subId = do
           ts_ <- TM.lookup tokenId (tokenSubscriptions st)
           forM_ ts_ $ \ts -> modifyTVar' ts $ S.delete subId
       )
+
+addTokenLastNtf :: NtfStore -> NtfTokenId -> PNMessageData -> IO (NonEmpty PNMessageData)
+addTokenLastNtf st tknId newNtf =
+  TM.lookupIO tknId (tokenLastNtfs st) >>= maybe (atomically maybeNewTokenLastNtfs) (atomically . addNtf)
+  where
+    maybeNewTokenLastNtfs =
+      TM.lookup tknId (tokenLastNtfs st) >>= maybe newTokenLastNtfs addNtf
+    newTokenLastNtfs = do
+      v <- newTVar [newNtf]
+      TM.insert tknId v $ tokenLastNtfs st
+      pure [newNtf]
+    addNtf v =
+      stateTVar v $ \ntfs -> let !ntfs' = rebuildList ntfs in (ntfs', ntfs')
+      where
+        rebuildList :: NonEmpty PNMessageData -> NonEmpty PNMessageData
+        rebuildList = foldr keepPrevNtf [newNtf]
+          where
+            PNMessageData {smpQueue = newNtfQ} = newNtf
+            keepPrevNtf ntf@PNMessageData {smpQueue} ntfs
+              | smpQueue /= newNtfQ && length ntfs < maxNtfs = ntf <| ntfs
+              | otherwise = ntfs
+        maxNtfs = 6
+
+-- This function is expected to be called after store log is read,
+-- as it checks for token existence when adding last notification.
+storeTokenLastNtf :: NtfStore -> NtfTokenId -> PNMessageData -> IO ()
+storeTokenLastNtf (NtfStore {tokens, tokenLastNtfs}) tknId ntf = do
+  TM.lookupIO tknId tokenLastNtfs >>= atomically . maybe newTokenLastNtfs (`modifyTVar'` (ntf <|))
+  where
+    newTokenLastNtfs = TM.lookup tknId tokenLastNtfs >>= maybe insertForExistingToken (`modifyTVar'` (ntf <|))
+    insertForExistingToken =
+      whenM (TM.member tknId tokens) $
+        TM.insertM tknId (newTVar [ntf]) tokenLastNtfs
+
+data TokenNtfMessageRecord = TNMRv1 NtfTokenId PNMessageData
+
+instance StrEncoding TokenNtfMessageRecord where
+  strEncode (TNMRv1 tknId ntf) = strEncode (Str "v1", tknId, ntf)
+  strP = "v1 " *> (TNMRv1 <$> strP_ <*> strP)

src/Simplex/Messaging/Notifications/Server/StoreLog.hs

@@ -37,8 +37,7 @@ import Simplex.Messaging.Notifications.Protocol
 import Simplex.Messaging.Notifications.Server.Store
 import Simplex.Messaging.Protocol (NtfPrivateAuthKey)
 import Simplex.Messaging.Server.StoreLog
-import Simplex.Messaging.Util (safeDecodeUtf8, whenM)
-import System.Directory (doesFileExist, renameFile)
+import Simplex.Messaging.Util (safeDecodeUtf8)
 import System.IO
 
 data NtfStoreLogRecord
@@ -50,6 +49,7 @@ data NtfStoreLogRecord
   | CreateSubscription NtfSubRec
   | SubscriptionStatus NtfSubscriptionId NtfSubStatus
   | DeleteSubscription NtfSubscriptionId
+  deriving (Show)
 
 data NtfTknRec = NtfTknRec
   { ntfTknId :: NtfTokenId,
@@ -61,6 +61,7 @@ data NtfTknRec = NtfTknRec
     tknRegCode :: NtfRegCode,
     tknCronInterval :: Word16
   }
+  deriving (Show)
 
 mkTknData :: NtfTknRec -> STM NtfTknData
 mkTknData NtfTknRec {ntfTknId, token, tknStatus = status, tknVerifyKey, tknDhKeys, tknDhSecret, tknRegCode, tknCronInterval = cronInt} = do
@@ -81,6 +82,7 @@ data NtfSubRec = NtfSubRec
     tokenId :: NtfTokenId,
     subStatus :: NtfSubStatus
   }
+  deriving (Show)
 
 mkSubData :: NtfSubRec -> STM NtfSubData
 mkSubData NtfSubRec {ntfSubId, smpQueue, notifierKey, tokenId, subStatus = status} = do
@@ -183,13 +185,7 @@ logDeleteSubscription :: StoreLog 'WriteMode -> NtfSubscriptionId -> IO ()
 logDeleteSubscription s subId = logNtfStoreRecord s $ DeleteSubscription subId
 
 readWriteNtfStore :: FilePath -> NtfStore -> IO (StoreLog 'WriteMode)
-readWriteNtfStore f st = do
-  whenM (doesFileExist f) $ do
-    readNtfStore f st
-    renameFile f $ f <> ".bak"
-  s <- openWriteStoreLog f
-  writeNtfStore s st
-  pure s
+readWriteNtfStore = readWriteStoreLog readNtfStore writeNtfStore
 
 readNtfStore :: FilePath -> NtfStore -> IO ()
 readNtfStore f st = mapM_ (addNtfLogRecord . LB.toStrict) . LB.lines =<< LB.readFile f

src/Simplex/Messaging/Notifications/Transport.hs

@@ -169,5 +169,6 @@ ntfTHandle c = THandle {connection = c, params}
           thServerVRange = versionToRange v,
           thAuth = Nothing,
           implySessId = False,
+          encryptBlock = Nothing,
           batch = False
         }

src/Simplex/Messaging/Notifications/Types.hs

@@ -11,7 +11,7 @@ import Data.Text.Encoding (decodeLatin1, encodeUtf8)
 import Data.Time (UTCTime)
 import Database.SQLite.Simple.FromField (FromField (..))
 import Database.SQLite.Simple.ToField (ToField (..))
-import Simplex.Messaging.Agent.Protocol (ConnId, NotificationsMode (..))
+import Simplex.Messaging.Agent.Protocol (ConnId, NotificationsMode (..), UserId)
 import qualified Simplex.Messaging.Crypto as C
 import Simplex.Messaging.Encoding
 import Simplex.Messaging.Notifications.Protocol
@@ -48,6 +48,7 @@ data NtfToken = NtfToken
     ntfServer :: NtfServer,
     ntfTokenId :: Maybe NtfTokenId,
     -- TODO combine keys to key pair as the types should match
+
     -- | key used by the ntf server to verify transmissions
     ntfPubKey :: C.APublicAuthKey,
     -- | key used by the ntf client to sign transmissions
@@ -79,17 +80,17 @@ newNtfToken deviceToken ntfServer (ntfPubKey, ntfPrivKey) ntfDhKeys ntfMode =
       ntfMode
     }
 
-data NtfSubAction = NtfSubNTFAction NtfSubNTFAction | NtfSubSMPAction NtfSubSMPAction
+data NtfSubAction = NSANtf NtfSubNTFAction | NSASMP NtfSubSMPAction
   deriving (Show)
 
 isDeleteNtfSubAction :: NtfSubAction -> Bool
 isDeleteNtfSubAction = \case
-  NtfSubNTFAction a -> case a of
+  NSANtf a -> case a of
     NSACreate -> False
     NSACheck -> False
     NSADelete -> True
     NSARotate -> True
-  NtfSubSMPAction a -> case a of
+  NSASMP a -> case a of
     NSASmpKey -> False
     NSASmpDelete -> True
 
@@ -98,8 +99,8 @@ type NtfActionTs = UTCTime
 data NtfSubNTFAction
   = NSACreate
   | NSACheck
-  | NSADelete
-  | NSARotate
+  | NSADelete -- deprecated
+  | NSARotate -- deprecated
   deriving (Show)
 
 instance Encoding NtfSubNTFAction where
@@ -177,7 +178,8 @@ instance FromField NtfAgentSubStatus where fromField = fromTextField_ $ either (
 instance ToField NtfAgentSubStatus where toField = toField . decodeLatin1 . smpEncode
 
 data NtfSubscription = NtfSubscription
-  { connId :: ConnId,
+  { userId :: UserId,
+    connId :: ConnId,
     smpServer :: SMPServer,
     ntfQueueId :: Maybe NotifierId,
     ntfServer :: NtfServer,
@@ -186,10 +188,11 @@ data NtfSubscription = NtfSubscription
   }
   deriving (Show)
 
-newNtfSubscription :: ConnId -> SMPServer -> Maybe NotifierId -> NtfServer -> NtfAgentSubStatus -> NtfSubscription
-newNtfSubscription connId smpServer ntfQueueId ntfServer ntfSubStatus =
+newNtfSubscription :: UserId -> ConnId -> SMPServer -> Maybe NotifierId -> NtfServer -> NtfAgentSubStatus -> NtfSubscription
+newNtfSubscription userId connId smpServer ntfQueueId ntfServer ntfSubStatus =
   NtfSubscription
-    { connId,
+    { userId,
+      connId,
       smpServer,
       ntfQueueId,
       ntfServer,

src/Simplex/Messaging/Protocol.hs

@@ -2,11 +2,13 @@
 {-# LANGUAGE BangPatterns #-}
 {-# LANGUAGE DataKinds #-}
 {-# LANGUAGE DeriveAnyClass #-}
+{-# LANGUAGE DerivingStrategies #-}
 {-# LANGUAGE DuplicateRecordFields #-}
 {-# LANGUAGE FlexibleContexts #-}
 {-# LANGUAGE FlexibleInstances #-}
 {-# LANGUAGE FunctionalDependencies #-}
 {-# LANGUAGE GADTs #-}
+{-# LANGUAGE GeneralizedNewtypeDeriving #-}
 {-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE NamedFieldPuns #-}
 {-# LANGUAGE OverloadedLists #-}
@@ -99,7 +101,8 @@ module Simplex.Messaging.Protocol
     BasicAuth (..),
     SrvLoc (..),
     CorrId (..),
-    EntityId,
+    EntityId (..),
+    pattern NoEntity,
     QueueId,
     RecipientId,
     SenderId,
@@ -255,13 +258,14 @@ supportedSMPClientVRange = mkVersionRange initialSMPClientVersion currentSMPClie
 -- TODO v6.0 remove dependency on version
 maxMessageLength :: VersionSMP -> Int
 maxMessageLength v
+  | v >= encryptedBlockSMPVersion = 16048 -- max 16051
   | v >= sendingProxySMPVersion = 16064 -- max 16067
-  | otherwise = 16088 -- 16064 - always use this size to determine allowed ranges
+  | otherwise = 16088 -- 16048 - always use this size to determine allowed ranges
 
 paddedProxiedTLength :: Int
-paddedProxiedTLength = 16242 -- 16241 .. 16243
+paddedProxiedTLength = 16226 -- 16225 .. 16227
 
--- TODO v6.0 change to 16064
+-- TODO v7.0 change to 16048
 type MaxMessageLen = 16088
 
 -- 16 extra bytes: 8 for timestamp and 8 for flags (7 flags and the space, only 1 flag is currently used)
@@ -269,10 +273,10 @@ type MaxRcvMessageLen = MaxMessageLen + 16 -- 16104, the padded size is 16106
 
 -- it is shorter to allow per-queue e2e encryption DH key in the "public" header
 e2eEncConfirmationLength :: Int
-e2eEncConfirmationLength = 15920 -- 15881 .. 15976
+e2eEncConfirmationLength = 15904 -- 15865 .. 15960
 
 e2eEncMessageLength :: Int
-e2eEncMessageLength = 16016 -- 16004 .. 16021
+e2eEncMessageLength = 16000 -- 15988 .. 16005
 
 -- | SMP protocol clients
 data Party = Recipient | Sender | Notifier | ProxiedClient
@@ -329,8 +333,8 @@ data RawTransmission = RawTransmission
   { authenticator :: ByteString, -- signature or encrypted transmission hash
     authorized :: ByteString, -- authorized transmission
     sessId :: SessionId,
-    corrId :: ByteString,
-    entityId :: ByteString,
+    corrId :: CorrId,
+    entityId :: EntityId,
     command :: ByteString
   }
   deriving (Show)
@@ -357,7 +361,7 @@ instance IsString (Maybe TransmissionAuth) where
   fromString = parseString $ B64.decode >=> C.decodeSignature >=> pure . fmap TASignature
 
 -- | unparsed sent SMP transmission with signature, without session ID.
-type SignedRawTransmission = (Maybe TransmissionAuth, SessionId, ByteString, ByteString)
+type SignedRawTransmission = (Maybe TransmissionAuth, CorrId, EntityId, ByteString)
 
 -- | unparsed sent SMP transmission with signature.
 type SentRawTransmission = (Maybe TransmissionAuth, ByteString)
@@ -374,7 +378,13 @@ type NotifierId = QueueId
 -- | SMP queue ID on the server.
 type QueueId = EntityId
 
-type EntityId = ByteString
+-- this type is used for server entities only
+newtype EntityId = EntityId {unEntityId :: ByteString}
+  deriving (Eq, Ord, Show)
+  deriving newtype (Encoding, StrEncoding)
+
+pattern NoEntity :: EntityId
+pattern NoEntity = EntityId ""
 
 -- | Parameterized type for SMP protocol commands from all clients.
 data Command (p :: Party) where
@@ -389,8 +399,6 @@ data Command (p :: Party) where
   NKEY :: NtfPublicAuthKey -> RcvNtfPublicDhKey -> Command Recipient
   NDEL :: Command Recipient
   GET :: Command Recipient
-  -- ACK v1 has to be supported for encoding/decoding
-  -- ACK :: Command Recipient
   ACK :: MsgId -> Command Recipient
   OFF :: Command Recipient
   DEL :: Command Recipient
@@ -475,6 +483,7 @@ data BrokerMsg where
   RRES :: EncFwdResponse -> BrokerMsg -- relay to proxy
   PRES :: EncResponse -> BrokerMsg -- proxy to client
   END :: BrokerMsg
+  DELD :: BrokerMsg
   INFO :: QueueInfo -> BrokerMsg
   OK :: BrokerMsg
   ERR :: ErrorType -> BrokerMsg
@@ -529,11 +538,13 @@ messageId :: Message -> MsgId
 messageId = \case
   Message {msgId} -> msgId
   MessageQuota {msgId} -> msgId
+{-# INLINE messageId #-}
 
 messageTs :: Message -> SystemTime
 messageTs = \case
   Message {msgTs} -> msgTs
   MessageQuota {msgTs} -> msgTs
+{-# INLINE messageTs #-}
 
 newtype EncRcvMsgBody = EncRcvMsgBody ByteString
   deriving (Eq, Show)
@@ -696,6 +707,7 @@ data BrokerMsgTag
   | RRES_
   | PRES_
   | END_
+  | DELD_
   | INFO_
   | OK_
   | ERR_
@@ -769,6 +781,7 @@ instance Encoding BrokerMsgTag where
     RRES_ -> "RRES"
     PRES_ -> "PRES"
     END_ -> "END"
+    DELD_ -> "DELD"
     INFO_ -> "INFO"
     OK_ -> "OK"
     ERR_ -> "ERR"
@@ -785,6 +798,7 @@ instance ProtocolMsgTag BrokerMsgTag where
     "RRES" -> Just RRES_
     "PRES" -> Just PRES_
     "END" -> Just END_
+    "DELD" -> Just DELD_
     "INFO" -> Just INFO_
     "OK" -> Just OK_
     "ERR" -> Just ERR_
@@ -1097,10 +1111,13 @@ serverStrP = do
     portP = show <$> (A.char ':' *> (A.decimal :: Parser Int))
 
 -- | Transmission correlation ID.
-newtype CorrId = CorrId {bs :: ByteString} deriving (Eq, Ord, Show)
+newtype CorrId = CorrId {bs :: ByteString}
+  deriving (Eq, Ord, Show)
+  deriving newtype (Encoding)
 
 instance IsString CorrId where
   fromString = CorrId . fromString
+  {-# INLINE fromString #-}
 
 instance StrEncoding CorrId where
   strEncode (CorrId cId) = strEncode cId
@@ -1181,6 +1198,8 @@ data ErrorType
     CRYPTO
   | -- | SMP queue capacity is exceeded on the server
     QUOTA
+  | -- | SMP server storage error
+    STORE {storeErr :: String}
   | -- | ACK command is sent without message to be acknowledged
     NO_MSG
   | -- | sent message is too large (> maxMessageLength = 16088 bytes)
@@ -1262,7 +1281,7 @@ transmissionP THandleParams {sessionId, implySessId} = do
       command <- A.takeByteString
       pure RawTransmission {authenticator, authorized = authorized', sessId, corrId, entityId, command}
 
-class (ProtocolEncoding v err msg, ProtocolEncoding v err (ProtoCommand msg), Show err, Show msg) => Protocol v err msg | msg -> v, msg -> err where
+class (ProtocolTypeI (ProtoType msg), ProtocolEncoding v err msg, ProtocolEncoding v err (ProtoCommand msg), Show err, Show msg) => Protocol v err msg | msg -> v, msg -> err where
   type ProtoCommand msg = cmd | cmd -> msg
   type ProtoType msg = (sch :: ProtocolType) | sch -> msg
   protocolClientHandshake :: forall c. Transport c => c -> Maybe C.KeyPairX25519 -> C.KeyHash -> VersionRange v -> ExceptT TransportError IO (THandle v c 'TClient)
@@ -1323,7 +1342,7 @@ instance PartyI p => ProtocolEncoding SMPVersion ErrorType (Command p) where
   fromProtocolError = fromProtocolError @SMPVersion @ErrorType @BrokerMsg
   {-# INLINE fromProtocolError #-}
 
-  checkCredentials (auth, _, entId, _) cmd = case cmd of
+  checkCredentials (auth, _, EntityId entId, _) cmd = case cmd of
     -- NEW must have signature but NOT queue ID
     NEW {}
       | isNothing auth -> Left $ CMD NO_AUTH
@@ -1411,6 +1430,9 @@ instance ProtocolEncoding SMPVersion ErrorType BrokerMsg where
     RRES (EncFwdResponse encBlock) -> e (RRES_, ' ', Tail encBlock)
     PRES (EncResponse encBlock) -> e (PRES_, ' ', Tail encBlock)
     END -> e END_
+    DELD
+      | v >= deletedEventSMPVersion -> e DELD_
+      | otherwise -> e END_
     INFO info -> e (INFO_, ' ', info)
     OK -> e OK_
     ERR err -> e (ERR_, ' ', err)
@@ -1436,6 +1458,7 @@ instance ProtocolEncoding SMPVersion ErrorType BrokerMsg where
     RRES_ -> RRES <$> (EncFwdResponse . unTail <$> _smpP)
     PRES_ -> PRES <$> (EncResponse . unTail <$> _smpP)
     END_ -> pure END
+    DELD_ -> pure DELD
     INFO_ -> INFO <$> _smpP
     OK_ -> pure OK
     ERR_ -> ERR <$> _smpP
@@ -1448,7 +1471,7 @@ instance ProtocolEncoding SMPVersion ErrorType BrokerMsg where
     PEBlock -> BLOCK
   {-# INLINE fromProtocolError #-}
 
-  checkCredentials (_, _, entId, _) cmd = case cmd of
+  checkCredentials (_, _, EntityId entId, _) cmd = case cmd of
     -- IDS response should not have queue ID
     IDS _ -> Right cmd
     -- ERR response does not always have queue ID
@@ -1492,6 +1515,7 @@ instance Encoding ErrorType where
     AUTH -> "AUTH"
     CRYPTO -> "CRYPTO"
     QUOTA -> "QUOTA"
+    STORE err -> "STORE " <> smpEncode err
     EXPIRED -> "EXPIRED"
     NO_MSG -> "NO_MSG"
     LARGE_MSG -> "LARGE_MSG"
@@ -1507,6 +1531,7 @@ instance Encoding ErrorType where
       "AUTH" -> pure AUTH
       "CRYPTO" -> pure CRYPTO
       "QUOTA" -> pure QUOTA
+      "STORE" -> STORE <$> _smpP
       "EXPIRED" -> pure EXPIRED
       "NO_MSG" -> pure NO_MSG
       "LARGE_MSG" -> pure LARGE_MSG
@@ -1643,8 +1668,8 @@ batchTransmissions' batch bSize ts
 batchTransmissions_ :: Int -> NonEmpty (Either TransportError ByteString, r) -> [TransportBatch r]
 batchTransmissions_ bSize = addBatch . foldr addTransmission ([], 0, 0, [], [])
   where
-    -- 3 = 2 bytes reserved for pad size + 1 for transmission count
-    bSize' = bSize - 3
+    -- 19 = 2 bytes reserved for pad size + 1 for transmission count + 16 auth tag from block encryption
+    bSize' = bSize - 19
     addTransmission :: (Either TransportError ByteString, r) -> ([TransportBatch r], Int, Int, [ByteString], [r]) -> ([TransportBatch r], Int, Int, [ByteString], [r])
     addTransmission (t_, r) acc@(bs, !len, !n, ss, rs) = case t_ of
       Left e -> (TBError e r : addBatch acc, 0, 0, [], [])
@@ -1721,16 +1746,16 @@ tDecodeParseValidate THandleParams {sessionId, thVersion = v, implySessId} = \ca
     | implySessId || sessId == sessionId ->
         let decodedTransmission = (,corrId,entityId,command) <$> decodeTAuthBytes authenticator
          in either (const $ tError corrId) (tParseValidate authorized) decodedTransmission
-    | otherwise -> (Nothing, "", (CorrId corrId, "", Left $ fromProtocolError @v @err @cmd PESession))
+    | otherwise -> (Nothing, "", (corrId, NoEntity, Left $ fromProtocolError @v @err @cmd PESession))
   Left _ -> tError ""
   where
-    tError :: ByteString -> SignedTransmission err cmd
-    tError corrId = (Nothing, "", (CorrId corrId, "", Left $ fromProtocolError @v @err @cmd PEBlock))
+    tError :: CorrId -> SignedTransmission err cmd
+    tError corrId = (Nothing, "", (corrId, NoEntity, Left $ fromProtocolError @v @err @cmd PEBlock))
 
     tParseValidate :: ByteString -> SignedRawTransmission -> SignedTransmission err cmd
     tParseValidate signed t@(sig, corrId, entityId, command) =
       let cmd = parseProtocol @v @err @cmd v command >>= checkCredentials t
-       in (sig, signed, (CorrId corrId, entityId, cmd))
+       in (sig, signed, (corrId, entityId, cmd))
 
 $(J.deriveJSON defaultJSON ''MsgFlags)
 

src/Simplex/Messaging/Server/CLI.hs

@@ -6,6 +6,8 @@
 {-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE NamedFieldPuns #-}
 {-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE ScopedTypeVariables #-}
+{-# LANGUAGE TupleSections #-}
 {-# LANGUAGE TypeApplications #-}
 
 module Simplex.Messaging.Server.CLI where
@@ -16,6 +18,7 @@ import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Char8 as B
 import Data.Either (fromRight)
 import Data.Ini (Ini, lookupValue)
+import Data.List ((\\))
 import Data.Text (Text)
 import qualified Data.Text as T
 import Data.Text.Encoding (encodeUtf8)
@@ -27,7 +30,7 @@ import Options.Applicative
 import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Protocol (ProtoServerWithAuth (..), ProtocolServer (..), ProtocolTypeI)
 import Simplex.Messaging.Transport (ATransport (..), TLS, Transport (..))
-import Simplex.Messaging.Transport.Server (loadFingerprint)
+import Simplex.Messaging.Transport.Server (AddHTTP, loadFileFingerprint)
 import Simplex.Messaging.Transport.WebSockets (WS)
 import Simplex.Messaging.Util (eitherToMaybe, whenM)
 import System.Directory (doesDirectoryExist, listDirectory, removeDirectoryRecursive, removePathForcibly)
@@ -41,11 +44,11 @@ import Text.Read (readMaybe)
 exitError :: String -> IO a
 exitError msg = putStrLn msg >> exitFailure
 
-confirmOrExit :: String -> IO ()
-confirmOrExit s =
+confirmOrExit :: String -> String -> IO ()
+confirmOrExit s no =
   withPrompt (s <> "\nContinue (Y/n): ") $ do
     ok <- getLine
-    when (ok /= "Y") $ putStrLn "Server NOT deleted" >> exitFailure
+    when (ok /= "Y") $ putStrLn no >> exitFailure
 
 data SignAlgorithm = ED448 | ED25519
   deriving (Read, Show)
@@ -137,7 +140,7 @@ createServerX509_ createCA cfgPath x509cfg = do
         )
 
     saveFingerprint = do
-      Fingerprint fp <- loadFingerprint $ c caCrtFile
+      Fingerprint fp <- loadFileFingerprint $ c caCrtFile
       withFile (c fingerprintFile) WriteMode (`B.hPutStrLn` strEncode fp)
       pure fp
 
@@ -266,26 +269,37 @@ settingIsOn section name ini
 checkSavedFingerprint :: FilePath -> X509Config -> IO ByteString
 checkSavedFingerprint cfgPath x509cfg = do
   savedFingerprint <- withFile (c fingerprintFile) ReadMode hGetLine
-  Fingerprint fp <- loadFingerprint (c caCrtFile)
+  Fingerprint fp <- loadFileFingerprint (c caCrtFile)
   when (B.pack savedFingerprint /= strEncode fp) $
     exitError "Stored fingerprint is invalid."
   pure fp
   where
     c = combine cfgPath . ($ x509cfg)
 
-iniTransports :: Ini -> [(String, ATransport)]
+iniTransports :: Ini -> [(ServiceName, ATransport, AddHTTP)]
 iniTransports ini =
-  let port = T.unpack $ strictIni "TRANSPORT" "port" ini
-      enableWebsockets = (== "on") $ strictIni "TRANSPORT" "websockets" ini
-   in (port, transport @TLS) : [("80", transport @WS) | enableWebsockets]
+  let smpPorts = ports $ strictIni "TRANSPORT" "port" ini
+      ws = strictIni "TRANSPORT" "websockets" ini
+      wsPorts
+        | ws == "off" = []
+        | ws == "on" = ["80"]
+        | otherwise = ports ws \\ smpPorts
+   in ts (transport @TLS) smpPorts <> ts (transport @WS) wsPorts
+  where
+    ts :: ATransport -> [ServiceName] -> [(ServiceName, ATransport, AddHTTP)]
+    ts t = map (\port -> (port, t, webPort == Just port))
+    webPort = T.unpack <$> eitherToMaybe (lookupValue "WEB" "https" ini)
+    ports = map T.unpack . T.splitOn ","
 
-printServerConfig :: [(ServiceName, ATransport)] -> Maybe FilePath -> IO ()
+printServerConfig :: [(ServiceName, ATransport, AddHTTP)] -> Maybe FilePath -> IO ()
 printServerConfig transports logFile = do
   putStrLn $ case logFile of
     Just f -> "Store log: " <> f
     _ -> "Store log disabled."
-  forM_ transports $ \(p, ATransport t) ->
-    putStrLn $ "Listening on port " <> p <> " (" <> transportName t <> ")..."
+  forM_ transports $ \(p, ATransport t, addHTTP) -> do
+    let descr = p <> " (" <> transportName t <> ")..."
+    putStrLn $ "Serving SMP protocol on port " <> descr
+    when addHTTP $ putStrLn $ "Serving static site on port " <> descr
 
 deleteDirIfExists :: FilePath -> IO ()
 deleteDirIfExists path = whenM (doesDirectoryExist path) $ removeDirectoryRecursive path

src/Simplex/Messaging/Server/Control.hs

@@ -4,11 +4,11 @@
 module Simplex.Messaging.Server.Control where
 
 import qualified Data.Attoparsec.ByteString.Char8 as A
-import Data.ByteString (ByteString)
 import Simplex.Messaging.Encoding.String
-import Simplex.Messaging.Protocol (BasicAuth)
+import Simplex.Messaging.Protocol (BasicAuth, SenderId)
 
 data CPClientRole = CPRNone | CPRUser | CPRAdmin
+  deriving (Eq)
 
 data ControlProtocol
   = CPAuth BasicAuth
@@ -20,7 +20,8 @@ data ControlProtocol
   | CPThreads
   | CPSockets
   | CPSocketThreads
-  | CPDelete ByteString
+  | CPServerInfo
+  | CPDelete SenderId
   | CPSave
   | CPHelp
   | CPQuit
@@ -37,6 +38,7 @@ instance StrEncoding ControlProtocol where
     CPThreads -> "threads"
     CPSockets -> "sockets"
     CPSocketThreads -> "socket-threads"
+    CPServerInfo -> "server-info"
     CPDelete bs -> "delete " <> strEncode bs
     CPSave -> "save"
     CPHelp -> "help"
@@ -53,6 +55,7 @@ instance StrEncoding ControlProtocol where
       "threads" -> pure CPThreads
       "sockets" -> pure CPSockets
       "socket-threads" -> pure CPSocketThreads
+      "server-info" -> pure CPServerInfo
       "delete" -> CPDelete <$> (A.space *> strP)
       "save" -> pure CPSave
       "help" -> pure CPHelp

src/Simplex/Messaging/Server/Env/STM.hs

@@ -1,24 +1,33 @@
 {-# LANGUAGE DataKinds #-}
 {-# LANGUAGE DuplicateRecordFields #-}
+{-# LANGUAGE FlexibleContexts #-}
+{-# LANGUAGE GADTs #-}
+{-# LANGUAGE KindSignatures #-}
+{-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE OverloadedStrings #-}
 {-# LANGUAGE ScopedTypeVariables #-}
 {-# LANGUAGE StrictData #-}
+{-# LANGUAGE TypeFamilies #-}
 
 module Simplex.Messaging.Server.Env.STM where
 
 import Control.Concurrent (ThreadId)
-import Control.Monad.IO.Unlift
+import Control.Logger.Simple
+import Control.Monad
+import qualified Crypto.PubKey.RSA as RSA
 import Crypto.Random
 import Data.ByteString.Char8 (ByteString)
 import Data.Int (Int64)
 import Data.IntMap.Strict (IntMap)
 import qualified Data.IntMap.Strict as IM
+import Data.List (intercalate)
 import Data.List.NonEmpty (NonEmpty)
-import Data.Map.Strict (Map)
-import qualified Data.Map.Strict as M
 import Data.Maybe (isJust, isNothing)
+import qualified Data.Text as T
 import Data.Time.Clock (getCurrentTime)
 import Data.Time.Clock.System (SystemTime)
+import qualified Data.X509 as X
 import Data.X509.Validation (Fingerprint (..))
 import Network.Socket (ServiceName)
 import qualified Network.TLS as T
@@ -30,29 +39,37 @@ import qualified Simplex.Messaging.Crypto as C
 import Simplex.Messaging.Protocol
 import Simplex.Messaging.Server.Expiration
 import Simplex.Messaging.Server.Information
+import Simplex.Messaging.Server.MsgStore.Journal
 import Simplex.Messaging.Server.MsgStore.STM
-import Simplex.Messaging.Server.QueueStore (NtfCreds (..), QueueRec (..))
+import Simplex.Messaging.Server.MsgStore.Types
+import Simplex.Messaging.Server.NtfStore
+import Simplex.Messaging.Server.QueueStore
 import Simplex.Messaging.Server.QueueStore.STM
 import Simplex.Messaging.Server.Stats
 import Simplex.Messaging.Server.StoreLog
 import Simplex.Messaging.TMap (TMap)
 import qualified Simplex.Messaging.TMap as TM
 import Simplex.Messaging.Transport (ATransport, VersionRangeSMP, VersionSMP)
-import Simplex.Messaging.Transport.Server (SocketState, TransportServerConfig, alpn, loadFingerprint, loadTLSServerParams, newSocketState)
+import Simplex.Messaging.Transport.Server
+import System.Directory (doesFileExist)
+import System.Exit (exitFailure)
 import System.IO (IOMode (..))
 import System.Mem.Weak (Weak)
 import UnliftIO.STM
 
 data ServerConfig = ServerConfig
-  { transports :: [(ServiceName, ATransport)],
+  { transports :: [(ServiceName, ATransport, AddHTTP)],
     smpHandshakeTimeout :: Int,
     tbqSize :: Natural,
-    -- serverTbqSize :: Natural,
+    msgStoreType :: AMSType,
     msgQueueQuota :: Int,
+    maxJournalMsgCount :: Int,
+    maxJournalStateLines :: Int,
     queueIdBytes :: Int,
     msgIdBytes :: Int,
     storeLogFile :: Maybe FilePath,
     storeMsgsFile :: Maybe FilePath,
+    storeNtfsFile :: Maybe FilePath,
     -- | set to False to prohibit creating new queues
     allowNewQueues :: Bool,
     -- | simple password that the clients need to pass in handshake to be able to create new queues
@@ -62,6 +79,11 @@ data ServerConfig = ServerConfig
     controlPortAdminAuth :: Maybe BasicAuth,
     -- | time after which the messages can be removed from the queues and check interval, seconds
     messageExpiration :: Maybe ExpirationConfig,
+    expireMessagesOnStart :: Bool,
+    -- | interval of inactivity after which journal queue is closed
+    idleQueueInterval :: Int64,
+    -- | notification expiration interval (seconds)
+    notificationExpiration :: ExpirationConfig,
     -- | time after which the socket with inactive client can be disconnected (without any messages or commands, incl. PING),
     -- and check interval, seconds
     inactiveClientExpiration :: Maybe ExpirationConfig,
@@ -74,10 +96,15 @@ data ServerConfig = ServerConfig
     serverStatsLogFile :: FilePath,
     -- | file to save and restore stats
     serverStatsBackupFile :: Maybe FilePath,
-    -- | CA certificate private key is not needed for initialization
-    caCertificateFile :: FilePath,
-    privateKeyFile :: FilePath,
-    certificateFile :: FilePath,
+    -- | interval and file to save prometheus metrics
+    prometheusInterval :: Maybe Int,
+    prometheusMetricsFile :: FilePath,
+    -- | notification delivery interval
+    ntfDeliveryInterval :: Int,
+    -- | interval between sending pending END events to unsubscribed clients, seconds
+    pendingENDInterval :: Int,
+    smpCredentials :: ServerCredentials,
+    httpCredentials :: Maybe ServerCredentials,
     -- | SMP client-server protocol version range
     smpServerVRange :: VersionRangeSMP,
     -- | TCP transport config
@@ -99,41 +126,90 @@ defaultMessageExpiration :: ExpirationConfig
 defaultMessageExpiration =
   ExpirationConfig
     { ttl = defMsgExpirationDays * 86400, -- seconds
-      checkInterval = 43200 -- seconds, 12 hours
+      checkInterval = 14400 -- seconds, 4 hours
+    }
+
+defaultIdleQueueInterval :: Int64
+defaultIdleQueueInterval = 28800 -- seconds, 8 hours
+
+defNtfExpirationHours :: Int64
+defNtfExpirationHours = 24
+
+defaultNtfExpiration :: ExpirationConfig
+defaultNtfExpiration =
+  ExpirationConfig
+    { ttl = defNtfExpirationHours * 3600, -- seconds
+      checkInterval = 3600 -- seconds, 1 hour
     }
 
 defaultInactiveClientExpiration :: ExpirationConfig
 defaultInactiveClientExpiration =
   ExpirationConfig
-    { ttl = 43200, -- seconds, 12 hours
+    { ttl = 21600, -- seconds, 6 hours
       checkInterval = 3600 -- seconds, 1 hours
     }
 
 defaultProxyClientConcurrency :: Int
 defaultProxyClientConcurrency = 32
 
+journalMsgStoreDepth :: Int
+journalMsgStoreDepth = 5
+
+defaultMaxJournalStateLines :: Int
+defaultMaxJournalStateLines = 16
+
+defaultMaxJournalMsgCount :: Int
+defaultMaxJournalMsgCount = 256
+
+defaultMsgQueueQuota :: Int
+defaultMsgQueueQuota = 128
+
+defaultStateTailSize :: Int
+defaultStateTailSize = 512
+
 data Env = Env
   { config :: ServerConfig,
+    serverActive :: TVar Bool,
     serverInfo :: ServerInformation,
     server :: Server,
     serverIdentity :: KeyHash,
-    queueStore :: QueueStore,
-    msgStore :: STMMsgStore,
+    msgStore :: AMsgStore,
+    ntfStore :: NtfStore,
     random :: TVar ChaChaDRG,
-    storeLog :: Maybe (StoreLog 'WriteMode),
-    tlsServerParams :: T.ServerParams,
+    tlsServerCreds :: T.Credential,
+    httpServerCreds :: Maybe T.Credential,
     serverStats :: ServerStats,
-    sockets :: SocketState,
+    sockets :: TVar [(ServiceName, SocketState)],
     clientSeq :: TVar ClientId,
-    clients :: TVar (IntMap Client),
+    clients :: TVar (IntMap (Maybe AClient)),
     proxyAgent :: ProxyAgent -- senders served on this proxy
   }
 
+type family MsgStore s where
+  MsgStore 'MSMemory = STMMsgStore
+  MsgStore 'MSJournal = JournalMsgStore
+
+data AMsgStore = forall s. (STMQueueStore (MsgStore s), MsgStoreClass (MsgStore s)) => AMS (SMSType s) (MsgStore s)
+
+data AStoreQueue = forall s. MsgStoreClass (MsgStore s) => ASQ (SMSType s) (StoreQueue (MsgStore s))
+
+data AMsgStoreCfg = forall s. MsgStoreClass (MsgStore s) => AMSC (SMSType s) (MsgStoreConfig (MsgStore s))
+
+msgPersistence :: AMsgStoreCfg -> Bool
+msgPersistence (AMSC SMSMemory (STMStoreConfig {storePath})) = isJust storePath
+msgPersistence (AMSC SMSJournal _) = True
+
+type Subscribed = Bool
+
 data Server = Server
-  { subscribedQ :: TQueue (RecipientId, Client),
-    subscribers :: TMap RecipientId Client,
-    ntfSubscribedQ :: TQueue (NotifierId, Client),
-    notifiers :: TMap NotifierId Client,
+  { subscribedQ :: TQueue (RecipientId, ClientId, Subscribed),
+    subscribers :: TMap RecipientId (TVar AClient),
+    ntfSubscribedQ :: TQueue (NotifierId, ClientId, Subscribed),
+    notifiers :: TMap NotifierId (TVar AClient),
+    subClients :: TVar (IntMap AClient), -- clients with SMP subscriptions
+    ntfSubClients :: TVar (IntMap AClient), -- clients with Ntf subscriptions
+    pendingSubEvents :: TVar (IntMap (NonEmpty (RecipientId, Subscribed))),
+    pendingNtfSubEvents :: TVar (IntMap (NonEmpty (NotifierId, Subscribed))),
     savingLock :: Lock
   }
 
@@ -143,11 +219,16 @@ newtype ProxyAgent = ProxyAgent
 
 type ClientId = Int
 
-data Client = Client
+data AClient = forall s. MsgStoreClass (MsgStore s) => AClient (SMSType s) (Client (MsgStore s))
+
+clientId' :: AClient -> ClientId
+clientId' (AClient _ Client {clientId}) = clientId
+
+data Client s = Client
   { clientId :: ClientId,
-    subscriptions :: TMap RecipientId (TVar Sub),
+    subscriptions :: TMap RecipientId Sub,
     ntfSubscriptions :: TMap NotifierId (),
-    rcvQ :: TBQueue (NonEmpty (Maybe QueueRec, Transmission Cmd)),
+    rcvQ :: TBQueue (NonEmpty (Maybe (StoreQueue s, QueueRec), Transmission Cmd)),
     sndQ :: TBQueue (NonEmpty (Transmission BrokerMsg)),
     msgQ :: TBQueue (NonEmpty (Transmission BrokerMsg)),
     procThreads :: TVar Int,
@@ -161,74 +242,103 @@ data Client = Client
     sndActiveAt :: TVar SystemTime
   }
 
-data SubscriptionThread = NoSub | SubPending | SubThread (Weak ThreadId) | ProhibitSub
+data ServerSub = ServerSub (TVar SubscriptionThread) | ProhibitSub
+
+data SubscriptionThread = NoSub | SubPending | SubThread (Weak ThreadId)
 
 data Sub = Sub
-  { subThread :: SubscriptionThread,
+  { subThread :: ServerSub, -- Nothing value indicates that sub
     delivered :: TMVar MsgId
   }
 
-newServer :: STM Server
+newServer :: IO Server
 newServer = do
-  subscribedQ <- newTQueue
-  subscribers <- TM.empty
-  ntfSubscribedQ <- newTQueue
-  notifiers <- TM.empty
-  savingLock <- createLock
-  return Server {subscribedQ, subscribers, ntfSubscribedQ, notifiers, savingLock}
+  subscribedQ <- newTQueueIO
+  subscribers <- TM.emptyIO
+  ntfSubscribedQ <- newTQueueIO
+  notifiers <- TM.emptyIO
+  subClients <- newTVarIO IM.empty
+  ntfSubClients <- newTVarIO IM.empty
+  pendingSubEvents <- newTVarIO IM.empty
+  pendingNtfSubEvents <- newTVarIO IM.empty
+  savingLock <- createLockIO
+  return Server {subscribedQ, subscribers, ntfSubscribedQ, notifiers, subClients, ntfSubClients, pendingSubEvents, pendingNtfSubEvents, savingLock}
 
-newClient :: TVar ClientId -> Natural -> VersionSMP -> ByteString -> SystemTime -> STM Client
-newClient nextClientId qSize thVersion sessionId createdAt = do
-  clientId <- stateTVar nextClientId $ \next -> (next, next + 1)
-  subscriptions <- TM.empty
-  ntfSubscriptions <- TM.empty
-  rcvQ <- newTBQueue qSize
-  sndQ <- newTBQueue qSize
-  msgQ <- newTBQueue qSize
-  procThreads <- newTVar 0
-  endThreads <- newTVar IM.empty
-  endThreadSeq <- newTVar 0
-  connected <- newTVar True
-  rcvActiveAt <- newTVar createdAt
-  sndActiveAt <- newTVar createdAt
+newClient :: SMSType s -> ClientId -> Natural -> VersionSMP -> ByteString -> SystemTime -> IO (Client (MsgStore s))
+newClient _msType clientId qSize thVersion sessionId createdAt = do
+  subscriptions <- TM.emptyIO
+  ntfSubscriptions <- TM.emptyIO
+  rcvQ <- newTBQueueIO qSize
+  sndQ <- newTBQueueIO qSize
+  msgQ <- newTBQueueIO qSize
+  procThreads <- newTVarIO 0
+  endThreads <- newTVarIO IM.empty
+  endThreadSeq <- newTVarIO 0
+  connected <- newTVarIO True
+  rcvActiveAt <- newTVarIO createdAt
+  sndActiveAt <- newTVarIO createdAt
   return Client {clientId, subscriptions, ntfSubscriptions, rcvQ, sndQ, msgQ, procThreads, endThreads, endThreadSeq, thVersion, sessionId, connected, createdAt, rcvActiveAt, sndActiveAt}
 
 newSubscription :: SubscriptionThread -> STM Sub
-newSubscription subThread = do
+newSubscription st = do
   delivered <- newEmptyTMVar
+  subThread <- ServerSub <$> newTVar st
   return Sub {subThread, delivered}
 
+newProhibitedSub :: STM Sub
+newProhibitedSub = do
+  delivered <- newEmptyTMVar
+  return Sub {subThread = ProhibitSub, delivered}
+
 newEnv :: ServerConfig -> IO Env
-newEnv config@ServerConfig {caCertificateFile, certificateFile, privateKeyFile, storeLogFile, smpAgentCfg, transportConfig, information, messageExpiration} = do
-  server <- atomically newServer
-  queueStore <- atomically newQueueStore
-  msgStore <- atomically newMsgStore
-  random <- liftIO C.newRandom
-  storeLog <- restoreQueues queueStore `mapM` storeLogFile
-  tlsServerParams <- loadTLSServerParams caCertificateFile certificateFile privateKeyFile (alpn transportConfig)
-  Fingerprint fp <- loadFingerprint caCertificateFile
+newEnv config@ServerConfig {smpCredentials, httpCredentials, storeLogFile, msgStoreType, storeMsgsFile, smpAgentCfg, information, messageExpiration, idleQueueInterval, msgQueueQuota, maxJournalMsgCount, maxJournalStateLines} = do
+  serverActive <- newTVarIO True
+  server <- newServer
+  msgStore@(AMS _ store) <- case msgStoreType of
+    AMSType SMSMemory -> AMS SMSMemory <$> newMsgStore STMStoreConfig {storePath = storeMsgsFile, quota = msgQueueQuota}
+    AMSType SMSJournal -> case storeMsgsFile of
+      Just storePath -> 
+        let cfg = JournalStoreConfig {storePath, quota = msgQueueQuota, pathParts = journalMsgStoreDepth, maxMsgCount = maxJournalMsgCount, maxStateLines = maxJournalStateLines, stateTailSize = defaultStateTailSize, idleInterval = idleQueueInterval}
+         in AMS SMSJournal <$> newMsgStore cfg
+      Nothing -> putStrLn "Error: journal msg store require path in [STORE_LOG], restore_messages" >> exitFailure
+  ntfStore <- NtfStore <$> TM.emptyIO
+  random <- C.newRandom
+  forM_ storeLogFile $ \f -> do
+    logInfo $ "restoring queues from file " <> T.pack f
+    sl <- readWriteQueueStore f store
+    setStoreLog store sl
+  tlsServerCreds <- getCredentials "SMP" smpCredentials
+  httpServerCreds <- mapM (getCredentials "HTTPS") httpCredentials
+  mapM_ checkHTTPSCredentials httpServerCreds
+  Fingerprint fp <- loadFingerprint smpCredentials
   let serverIdentity = KeyHash fp
-  serverStats <- atomically . newServerStats =<< getCurrentTime
-  sockets <- atomically newSocketState
+  serverStats <- newServerStats =<< getCurrentTime
+  sockets <- newTVarIO []
   clientSeq <- newTVarIO 0
   clients <- newTVarIO mempty
-  proxyAgent <- atomically $ newSMPProxyAgent smpAgentCfg random
-  pure Env {config, serverInfo, server, serverIdentity, queueStore, msgStore, random, storeLog, tlsServerParams, serverStats, sockets, clientSeq, clients, proxyAgent}
+  proxyAgent <- newSMPProxyAgent smpAgentCfg random
+  pure Env {serverActive, config, serverInfo, server, serverIdentity, msgStore, ntfStore, random, tlsServerCreds, httpServerCreds, serverStats, sockets, clientSeq, clients, proxyAgent}
   where
-    restoreQueues :: QueueStore -> FilePath -> IO (StoreLog 'WriteMode)
-    restoreQueues QueueStore {queues, senders, notifiers} f = do
-      (qs, s) <- readWriteStoreLog f
-      atomically $ do
-        writeTVar queues =<< mapM newTVar qs
-        writeTVar senders $! M.foldr' addSender M.empty qs
-        writeTVar notifiers $! M.foldr' addNotifier M.empty qs
-      pure s
-    addSender :: QueueRec -> Map SenderId RecipientId -> Map SenderId RecipientId
-    addSender q = M.insert (senderId q) (recipientId q)
-    addNotifier :: QueueRec -> Map NotifierId RecipientId -> Map NotifierId RecipientId
-    addNotifier q = case notifier q of
-      Nothing -> id
-      Just NtfCreds {notifierId} -> M.insert notifierId (recipientId q)
+    getCredentials protocol creds = do
+      files <- missingCreds
+      unless (null files) $ do
+        putStrLn $ "Error: no " <> protocol <> " credentials: " <> intercalate ", " files
+        when (protocol == "HTTPS") $ putStrLn letsEncrypt
+        exitFailure
+      loadServerCredential creds
+      where
+        missingfile f = (\y -> [f | not y]) <$> doesFileExist f
+        missingCreds = do
+          let files = maybe id (:) (caCertificateFile creds) [certificateFile creds, privateKeyFile creds]
+           in concat <$> mapM missingfile files
+    checkHTTPSCredentials (X.CertificateChain cc, _k) =
+      -- LetsEncrypt provides ECDSA with insecure curve p256 (https://safecurves.cr.yp.to)
+      case map (X.signedObject . X.getSigned) cc of
+        X.Certificate {X.certPubKey = X.PubKeyRSA rsa} : _ca | RSA.public_size rsa >= 512 -> pure ()
+        _ -> do
+          putStrLn $ "Error: unsupported HTTPS credentials, required 4096-bit RSA\n" <> letsEncrypt
+          exitFailure
+    letsEncrypt = "Use Let's Encrypt to generate: certbot certonly --standalone -d yourdomainname --key-type rsa --rsa-key-size 4096"
     serverInfo =
       ServerInformation
         { information,
@@ -244,10 +354,13 @@ newEnv config@ServerConfig {caCertificateFile, certificateFile, privateKeyFile,
       where
         persistence
           | isNothing storeLogFile = SPMMemoryOnly
-          | isJust (storeMsgsFile config) = SPMMessages
+          | isJust storeMsgsFile = SPMMessages
           | otherwise = SPMQueues
 
-newSMPProxyAgent :: SMPClientAgentConfig -> TVar ChaChaDRG -> STM ProxyAgent
+newSMPProxyAgent :: SMPClientAgentConfig -> TVar ChaChaDRG -> IO ProxyAgent
 newSMPProxyAgent smpAgentCfg random = do
   smpAgent <- newSMPClientAgent smpAgentCfg random
   pure ProxyAgent {smpAgent}
+
+readWriteQueueStore :: STMQueueStore s => FilePath -> s -> IO (StoreLog 'WriteMode)
+readWriteQueueStore = readWriteStoreLog readQueueStore writeQueueStore

src/Simplex/Messaging/Server/Information.hs

@@ -1,14 +1,21 @@
 {-# LANGUAGE DataKinds #-}
 {-# LANGUAGE DuplicateRecordFields #-}
+{-# LANGUAGE LambdaCase #-}
+{-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE OverloadedStrings #-}
 {-# LANGUAGE StrictData #-}
 {-# LANGUAGE TemplateHaskell #-}
 
 module Simplex.Messaging.Server.Information where
 
+import Data.Aeson (FromJSON (..), ToJSON (..))
 import qualified Data.Aeson.TH as J
+import qualified Data.Attoparsec.ByteString.Char8 as A
 import Data.Int (Int64)
+import Data.Maybe (isJust)
 import Data.Text (Text)
 import Simplex.Messaging.Agent.Protocol (ConnectionMode (..), ConnectionRequestUri)
+import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Parsers (defaultJSON, dropPrefix, enumJSON)
 
 data ServerInformation = ServerInformation
@@ -36,16 +43,59 @@ data ServerPublicInfo = ServerPublicInfo
     adminContacts :: Maybe ServerContactAddress,
     complaintsContacts :: Maybe ServerContactAddress,
     hosting :: Maybe Entity,
+    hostingType :: Maybe HostingType,
     serverCountry :: Maybe Text
   }
   deriving (Show)
 
+emptyServerInfo :: Text -> ServerPublicInfo
+emptyServerInfo sourceCode =
+  ServerPublicInfo
+    { sourceCode,
+      usageConditions = Nothing,
+      operator = Nothing,
+      website = Nothing,
+      adminContacts = Nothing,
+      complaintsContacts = Nothing,
+      hosting = Nothing,
+      hostingType = Nothing,
+      serverCountry = Nothing
+    }
+
+hasServerInfo :: ServerPublicInfo -> Bool
+hasServerInfo ServerPublicInfo {usageConditions, operator, website, adminContacts, complaintsContacts, hosting, hostingType, serverCountry} =
+  isJust usageConditions || isJust operator || isJust website || isJust adminContacts || isJust complaintsContacts || isJust hosting || isJust hostingType || isJust serverCountry
+
 data ServerPersistenceMode = SPMMemoryOnly | SPMQueues | SPMMessages
   deriving (Show)
 
 data ServerConditions = ServerConditions {conditions :: Text, amendments :: Maybe Text}
   deriving (Show)
 
+data HostingType = HTVirtual | HTDedicated | HTColocation | HTOwned
+  deriving (Show)
+
+instance StrEncoding HostingType where
+  strEncode = \case
+    HTVirtual -> "virtual"
+    HTDedicated -> "dedicated"
+    HTColocation -> "colocation"
+    HTOwned -> "owned"
+  strP =
+    A.takeTill (== ' ') >>= \case
+      "virtual" -> pure HTVirtual
+      "dedicated" -> pure HTDedicated
+      "colocation" -> pure HTColocation
+      "owned" -> pure HTOwned
+      _ -> fail "bad HostingType"
+
+instance FromJSON HostingType where
+  parseJSON = strParseJSON "HostingType"
+
+instance ToJSON HostingType where
+  toJSON = strToJSON
+  toEncoding = strToJEncoding
+
 data Entity = Entity {name :: Text, country :: Maybe Text}
   deriving (Show)
 

src/Simplex/Messaging/Server/Main.hs

@@ -1,5 +1,6 @@
 {-# LANGUAGE ApplicativeDo #-}
 {-# LANGUAGE DuplicateRecordFields #-}
+{-# LANGUAGE GADTs #-}
 {-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE NamedFieldPuns #-}
 {-# LANGUAGE OverloadedLists #-}
@@ -8,15 +9,18 @@
 {-# LANGUAGE ScopedTypeVariables #-}
 {-# LANGUAGE StrictData #-}
 {-# LANGUAGE TypeApplications #-}
+{-# OPTIONS_GHC -fno-warn-ambiguous-fields #-}
 
 module Simplex.Messaging.Server.Main where
 
 import Control.Concurrent.STM
+import Control.Exception (finally)
 import Control.Logger.Simple
-import Control.Monad (void, when, (<$!>))
+import Control.Monad
 import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Char8 as B
 import Data.Char (isAlpha, isAscii, toUpper)
+import Data.Either (fromRight)
 import Data.Functor (($>))
 import Data.Ini (Ini, lookupValue, readIniFile)
 import Data.List (find, isPrefixOf)
@@ -24,68 +28,149 @@ import qualified Data.List.NonEmpty as L
 import Data.Maybe (fromMaybe, isJust, isNothing)
 import Data.Text (Text)
 import qualified Data.Text as T
-import Data.Text.Encoding (encodeUtf8)
+import Data.Text.Encoding (decodeLatin1, encodeUtf8)
 import qualified Data.Text.IO as T
 import Network.Socket (HostName)
 import Options.Applicative
 import Simplex.Messaging.Agent.Protocol (connReqUriP')
-import Simplex.Messaging.Client (HostMode (..), NetworkConfig (..), ProtocolClientConfig (..), SocksMode (..), defaultNetworkConfig)
+import Simplex.Messaging.Client (HostMode (..), NetworkConfig (..), ProtocolClientConfig (..), SocksMode (..), defaultNetworkConfig, textToHostMode)
 import Simplex.Messaging.Client.Agent (SMPClientAgentConfig (..), defaultSMPClientAgentConfig)
 import qualified Simplex.Messaging.Crypto as C
 import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Parsers (parseAll)
 import Simplex.Messaging.Protocol (BasicAuth (..), ProtoServerWithAuth (ProtoServerWithAuth), pattern SMPServer)
-import Simplex.Messaging.Server (runSMPServer)
+import Simplex.Messaging.Server (AttachHTTP, exportMessages, importMessages, printMessageStats, runSMPServer)
 import Simplex.Messaging.Server.CLI
-import Simplex.Messaging.Server.Env.STM (ServerConfig (..), defMsgExpirationDays, defaultInactiveClientExpiration, defaultMessageExpiration, defaultProxyClientConcurrency)
+import Simplex.Messaging.Server.Env.STM
 import Simplex.Messaging.Server.Expiration
 import Simplex.Messaging.Server.Information
-import Simplex.Messaging.Transport (batchCmdsSMPVersion, sendingProxySMPVersion, simplexMQVersion, supportedSMPHandshakes, supportedServerSMPRelayVRange)
-import Simplex.Messaging.Transport.Client (TransportHost (..))
-import Simplex.Messaging.Transport.Server (TransportServerConfig (..), defaultTransportServerConfig)
-import Simplex.Messaging.Util (eitherToMaybe, safeDecodeUtf8, tshow)
+import Simplex.Messaging.Server.MsgStore.Journal (JournalStoreConfig (..))
+import Simplex.Messaging.Server.MsgStore.Types (AMSType (..), SMSType (..), newMsgStore)
+import Simplex.Messaging.Server.QueueStore.STM (readQueueStore)
+import Simplex.Messaging.Transport (batchCmdsSMPVersion, sendingProxySMPVersion, simplexMQVersion, supportedServerSMPRelayVRange)
+import Simplex.Messaging.Transport.Client (SocksProxy, TransportHost (..), defaultSocksProxy)
+import Simplex.Messaging.Transport.Server (ServerCredentials (..), TransportServerConfig (..), defaultTransportServerConfig)
+import Simplex.Messaging.Util (eitherToMaybe, ifM, safeDecodeUtf8, tshow)
 import Simplex.Messaging.Version (mkVersionRange)
-import System.Directory (createDirectoryIfMissing, doesFileExist)
+import System.Directory (createDirectoryIfMissing, doesDirectoryExist, doesFileExist)
+import System.Exit (exitFailure)
 import System.FilePath (combine)
 import System.IO (BufferMode (..), hSetBuffering, stderr, stdout)
 import Text.Read (readMaybe)
 
 smpServerCLI :: FilePath -> FilePath -> IO ()
-smpServerCLI = smpServerCLI_ (\_ _ _ -> pure ()) (\_ -> pure ())
+smpServerCLI = smpServerCLI_ (\_ _ _ -> pure ()) (\_ -> pure ()) (\_ -> error "attachStaticFiles not available")
 
-smpServerCLI_ :: (ServerInformation -> Maybe TransportHost -> FilePath -> IO ()) -> (EmbeddedWebParams -> IO ()) -> FilePath -> FilePath -> IO ()
-smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
+smpServerCLI_ ::
+  (ServerInformation -> Maybe TransportHost -> FilePath -> IO ()) ->
+  (EmbeddedWebParams -> IO ()) ->
+  (FilePath -> (AttachHTTP -> IO ()) -> IO ()) ->
+  FilePath ->
+  FilePath ->
+  IO ()
+smpServerCLI_ generateSite serveStaticFiles attachStaticFiles cfgPath logPath =
   getCliCommand' (cliCommandP cfgPath logPath iniFile) serverVersion >>= \case
     Init opts ->
       doesFileExist iniFile >>= \case
         True -> exitError $ "Error: server is already initialized (" <> iniFile <> " exists).\nRun `" <> executableName <> " start`."
         _ -> initializeServer opts
-    OnlineCert certOpts ->
-      doesFileExist iniFile >>= \case
-        True -> genOnline cfgPath certOpts
-        _ -> exitError $ "Error: server is not initialized (" <> iniFile <> " does not exist).\nRun `" <> executableName <> " init`."
-    Start ->
-      doesFileExist iniFile >>= \case
-        True -> readIniFile iniFile >>= either exitError runServer
-        _ -> exitError $ "Error: server is not initialized (" <> iniFile <> " does not exist).\nRun `" <> executableName <> " init`."
+    OnlineCert certOpts -> withIniFile $ \_ -> genOnline cfgPath certOpts
+    Start -> withIniFile runServer
     Delete -> do
-      confirmOrExit "WARNING: deleting the server will make all queues inaccessible, because the server identity (certificate fingerprint) will change.\nTHIS CANNOT BE UNDONE!"
+      confirmOrExit
+        "WARNING: deleting the server will make all queues inaccessible, because the server identity (certificate fingerprint) will change.\nTHIS CANNOT BE UNDONE!"
+        "Server NOT deleted"
       deleteDirIfExists cfgPath
       deleteDirIfExists logPath
       putStrLn "Deleted configuration and log files"
+    Journal cmd -> withIniFile $ \ini -> do
+      msgsDirExists <- doesDirectoryExist storeMsgsJournalDir
+      msgsFileExists <- doesFileExist storeMsgsFilePath
+      let enableStoreLog = settingIsOn "STORE_LOG" "enable" ini
+      storeLogFile <- case enableStoreLog $> storeLogFilePath of
+        Just storeLogFile -> do
+          ifM
+            (doesFileExist storeLogFile)
+            (pure storeLogFile)
+            (putStrLn ("Store log file " <> storeLogFile <> " not found") >> exitFailure)
+        Nothing -> putStrLn "Store log disabled, see `[STORE_LOG] enable`" >> exitFailure
+      case cmd of
+        JCImport
+          | msgsFileExists && msgsDirExists -> exitConfigureMsgStorage
+          | msgsDirExists -> do
+              putStrLn $ storeMsgsJournalDir <> " directory already exists."
+              exitFailure
+          | not msgsFileExists -> do
+              putStrLn $ storeMsgsFilePath <> " file does not exists."
+              exitFailure
+          | otherwise -> do
+              confirmOrExit
+                ("WARNING: message log file " <> storeMsgsFilePath <> " will be imported to journal directory " <> storeMsgsJournalDir)
+                "Messages not imported"
+              ms <- newJournalMsgStore
+              readQueueStore storeLogFile ms
+              msgStats <- importMessages True ms storeMsgsFilePath Nothing -- no expiration
+              putStrLn "Import completed"
+              printMessageStats "Messages" msgStats
+              putStrLn $ case readMsgStoreType ini of
+                Right (AMSType SMSMemory) -> "store_messages set to `memory`, update it to `journal` in INI file"
+                Right (AMSType SMSJournal) -> "store_messages set to `journal`"
+                Left e -> e <> ", update it to `journal` in INI file"
+        JCExport
+          | msgsFileExists && msgsDirExists -> exitConfigureMsgStorage
+          | msgsFileExists -> do
+              putStrLn $ storeMsgsFilePath <> " file already exists."
+              exitFailure
+          | otherwise -> do
+              confirmOrExit
+                ("WARNING: journal directory " <> storeMsgsJournalDir <> " will be exported to message log file " <> storeMsgsFilePath)
+                "Journal not exported"
+              ms <- newJournalMsgStore
+              readQueueStore storeLogFile ms
+              exportMessages True ms storeMsgsFilePath False
+              putStrLn "Export completed"
+              putStrLn $ case readMsgStoreType ini of
+                Right (AMSType SMSMemory) -> "store_messages set to `memory`"
+                Right (AMSType SMSJournal) -> "store_messages set to `journal`, update it to `memory` in INI file"
+                Left e -> e <> ", update it to `memory` in INI file"
+        JCDelete
+          | not msgsDirExists -> do
+              putStrLn $ storeMsgsJournalDir <> " directory does not exists."
+              exitFailure
+          | otherwise -> do
+              confirmOrExit
+                ("WARNING: journal directory " <> storeMsgsJournalDir <> " will be permanently deleted.\nTHIS CANNOT BE UNDONE!")
+                "Messages NOT deleted"
+              deleteDirIfExists storeMsgsJournalDir
+              putStrLn $ "Deleted all messages in journal " <> storeMsgsJournalDir
   where
+    withIniFile a =
+      doesFileExist iniFile >>= \case
+        True -> readIniFile iniFile >>= either exitError a
+        _ -> exitError $ "Error: server is not initialized (" <> iniFile <> " does not exist).\nRun `" <> executableName <> " init`."
+    newJournalMsgStore = newMsgStore JournalStoreConfig {storePath = storeMsgsJournalDir, pathParts = journalMsgStoreDepth, quota = defaultMsgQueueQuota, maxMsgCount = defaultMaxJournalMsgCount, maxStateLines = defaultMaxJournalStateLines, stateTailSize = defaultStateTailSize, idleInterval = checkInterval defaultMessageExpiration}
     iniFile = combine cfgPath "smp-server.ini"
     serverVersion = "SMP server v" <> simplexMQVersion
-    defaultServerPort = "5223"
+    defaultServerPorts = "5223,443"
     executableName = "smp-server"
     storeLogFilePath = combine logPath "smp-server-store.log"
-    httpsCertFile = combine cfgPath "web.cert"
+    storeMsgsFilePath = combine logPath "smp-server-messages.log"
+    storeMsgsJournalDir = combine logPath "messages"
+    storeNtfsFilePath = combine logPath "smp-server-ntfs.log"
+    readMsgStoreType :: Ini -> Either String AMSType
+    readMsgStoreType = textToMsgStoreType . fromRight "memory" . lookupValue "STORE_LOG" "store_messages"
+    textToMsgStoreType = \case
+      "memory" -> Right $ AMSType SMSMemory
+      "journal" -> Right $ AMSType SMSJournal
+      s -> Left $ "invalid store_messages: " <> T.unpack s
+    httpsCertFile = combine cfgPath "web.crt"
     httpsKeyFile = combine cfgPath "web.key"
     defaultStaticPath = combine logPath "www"
     initializeServer opts@InitOptions {ip, fqdn, sourceCode = src', webStaticPath = sp', disableWeb = noWeb', scripted}
       | scripted = initialize opts
       | otherwise = do
           putStrLn "Use `smp-server init -h` for available options."
+          checkInitOptions opts
           void $ withPrompt "SMP server will be initialized (press Enter)" getLine
           enableStoreLog <- onOffPrompt "Enable store log to restore queues and messages on server restart" True
           logStats <- onOffPrompt "Enable logging daily statistics" False
@@ -95,16 +180,15 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
           host' <- withPrompt ("Enter server FQDN or IP address for certificate (" <> host <> "): ") getLine
           sourceCode' <- withPrompt ("Enter server source code URI (" <> maybe simplexmqSource T.unpack src' <> "): ") getServerSourceCode
           staticPath' <- withPrompt ("Enter path to store generated static site with server information (" <> fromMaybe defaultStaticPath sp' <> "): ") getLine
-          enableWeb <- onOffPrompt "Enable built-in web server for static site" (not noWeb')
           initialize
             opts
               { enableStoreLog,
                 logStats,
                 fqdn = if null host' then fqdn else Just host',
                 password,
-                sourceCode = (T.pack <$> sourceCode') <|> src',
+                sourceCode = (T.pack <$> sourceCode') <|> src' <|> Just (T.pack simplexmqSource),
                 webStaticPath = if null staticPath' then sp' else Just staticPath',
-                disableWeb = not enableWeb
+                disableWeb = noWeb'
               }
       where
         serverPassword =
@@ -116,7 +200,18 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
               case strDecode $ encodeUtf8 $ T.pack s of
                 Right auth -> pure . Just $ ServerPassword auth
                 _ -> putStrLn "Invalid password. Only latin letters, digits and symbols other than '@' and ':' are allowed" >> serverPassword
-        initialize InitOptions {enableStoreLog, logStats, signAlgorithm, password, sourceCode, webStaticPath, disableWeb} = do
+        checkInitOptions InitOptions {sourceCode, serverInfo, operatorCountry, hostingCountry} = do
+          let err_
+                | isNothing sourceCode && hasServerInfo serverInfo =
+                    Just "Error: passing any server information requires passing --source-code"
+                | isNothing (operator serverInfo) && isJust operatorCountry =
+                    Just "Error: passing --operator-country requires passing --operator"
+                | isNothing (hosting serverInfo) && isJust hostingCountry =
+                    Just "Error: passing --hosting-country requires passing --hosting"
+                | otherwise = Nothing
+          forM_ err_ $ \err -> putStrLn err >> exitFailure
+        initialize opts'@InitOptions {enableStoreLog, logStats, signAlgorithm, password, controlPort, socksProxy, ownDomains, sourceCode, webStaticPath, disableWeb} = do
+          checkInitOptions opts'
           clearDirIfExists cfgPath
           clearDirIfExists logPath
           createDirectoryIfMissing True cfgPath
@@ -124,9 +219,10 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
           let x509cfg = defaultX509Config {commonName = fromMaybe ip fqdn, signAlgorithm}
           fp <- createServerX509 cfgPath x509cfg
           basicAuth <- mapM createServerPassword password
+          controlPortPwds <- forM controlPort $ \_ -> let pwd = decodeLatin1 <$> randomBase64 18 in (,) <$> pwd <*> pwd
           let host = fromMaybe (if ip == "127.0.0.1" then "<hostnames>" else ip) fqdn
               srv = ProtoServerWithAuth (SMPServer [THDomainName host] "" (C.KeyHash fp)) basicAuth
-          T.writeFile iniFile $ iniFileContent host basicAuth $ Just "https://github.com/simplex-chat/simplexmq"
+          T.writeFile iniFile $ iniFileContent host basicAuth controlPortPwds
           putStrLn $ "Server initialized, please provide additional server information in " <> iniFile <> "."
           putStrLn $ "Run `" <> executableName <> " start` to start server."
           warnCAPrivateKeyFile cfgPath x509cfg
@@ -135,9 +231,10 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
           where
             createServerPassword = \case
               ServerPassword s -> pure s
-              SPRandom -> BasicAuth . strEncode <$> (atomically . C.randomBytes 32 =<< C.newRandom)
-            iniFileContent host basicAuth sourceCode' =
-              informationIniContent sourceCode'
+              SPRandom -> BasicAuth <$> randomBase64 32
+            randomBase64 n = strEncode <$> (atomically . C.randomBytes n =<< C.newRandom)
+            iniFileContent host basicAuth controlPortPwds =
+              informationIniContent opts'
                 <> "[STORE_LOG]\n\
                    \# The server uses STM memory for persistence,\n\
                    \# that will be lost on restart (e.g., as with redis).\n\
@@ -145,13 +242,20 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
                    \# and restoring it when the server is started.\n\
                    \# Log is compacted on start (deleted objects are removed).\n"
                 <> ("enable: " <> onOff enableStoreLog <> "\n\n")
-                <> "# Undelivered messages are optionally saved and restored when the server restarts,\n\
-                   \# they are preserved in the .bak file until the next restart.\n"
-                <> ("restore_messages: " <> onOff enableStoreLog <> "\n")
-                <> ("expire_messages_days: " <> tshow defMsgExpirationDays <> "\n\n")
+                <> "# Message storage mode: `memory` or `journal`.\n\
+                   \store_messages: memory\n\n\
+                   \# When store_messages is `memory`, undelivered messages are optionally saved and restored\n\
+                   \# when the server restarts, they are preserved in the .bak file until the next restart.\n"
+                <> ("restore_messages: " <> onOff enableStoreLog <> "\n\n")
+                <> "# Messages and notifications expiration periods.\n"
+                <> ("expire_messages_days: " <> tshow defMsgExpirationDays <> "\n")
+                <> "expire_messages_on_start: on\n"
+                <> ("expire_ntfs_hours: " <> tshow defNtfExpirationHours <> "\n\n")
                 <> "# Log daily server statistics to CSV file\n"
                 <> ("log_stats: " <> onOff logStats <> "\n\n")
-                <> "[AUTH]\n\
+                <> "# Log interval for real-time Prometheus metrics\n\
+                   \# prometheus_interval: 300\n\n\
+                   \[AUTH]\n\
                    \# Set new_queues option to off to completely prohibit creating new messaging queues.\n\
                    \# This can be useful when you want to decommission the server, but not all connections are switched yet.\n\
                    \new_queues: on\n\n\
@@ -160,66 +264,77 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
                    \# smp://fingerprint:password@host1,host2\n\
                    \# The password will not be shared with the connecting contacts, you must share it only\n\
                    \# with the users who you want to allow creating messaging queues on your server.\n"
-                <> ( case basicAuth of
-                      Just auth -> "create_password: " <> safeDecodeUtf8 (strEncode auth)
-                      _ -> "# create_password: password to create new queues (any printable ASCII characters without whitespace, '@', ':' and '/')"
+                <> ( let noPassword = "password to create new queues and forward messages (any printable ASCII characters without whitespace, '@', ':' and '/')"
+                      in optDisabled basicAuth <> "create_password: " <> maybe noPassword (safeDecodeUtf8 . strEncode) basicAuth
                    )
-                <> "\n\n\
-                   \# control_port_admin_password:\n\
-                   \# control_port_user_password:\n\n\
+                <> "\n\n"
+                <> (optDisabled controlPortPwds <> "control_port_admin_password: " <> maybe "" fst controlPortPwds <> "\n")
+                <> (optDisabled controlPortPwds <> "control_port_user_password: " <> maybe "" snd controlPortPwds <> "\n")
+                <> "\n\
                    \[TRANSPORT]\n\
-                   \# host is only used to print server address on start\n"
+                   \# Host is only used to print server address on start.\n\
+                   \# You can specify multiple server ports.\n"
                 <> ("host: " <> T.pack host <> "\n")
-                <> ("port: " <> T.pack defaultServerPort <> "\n")
-                <> "log_tls_errors: off\n\
-                   \websockets: off\n\
-                   \# control_port: 5224\n\n\
+                <> ("port: " <> T.pack defaultServerPorts <> "\n")
+                <> "log_tls_errors: off\n\n\
+                   \# Use `websockets: 443` to run websockets server in addition to plain TLS.\n\
+                   \# This option is deprecated and should be used for testing only.\n\
+                   \# , port 443 should be specified in port above\n\
+                   \websockets: off\n"
+                <> (optDisabled controlPort <> "control_port: " <> tshow (fromMaybe defaultControlPort controlPort))
+                <> "\n\n\
                    \[PROXY]\n\
                    \# Network configuration for SMP proxy client.\n\
                    \# `host_mode` can be 'public' (default) or 'onion'.\n\
                    \# It defines prefferred hostname for destination servers with multiple hostnames.\n\
                    \# host_mode: public\n\
                    \# required_host_mode: off\n\n\
-                   \# The domain suffixes of the relays you operate (space-separated) to count as separate proxy statistics.\n\
-                   \# own_server_domains: \n\n\
+                   \# The domain suffixes of the relays you operate (space-separated) to count as separate proxy statistics.\n"
+                <> (optDisabled ownDomains <> "own_server_domains: " <> maybe "" (safeDecodeUtf8 . strEncode) ownDomains)
+                <> "\n\n\
                    \# SOCKS proxy port for forwarding messages to destination servers.\n\
-                   \# You may need a separate instance of SOCKS proxy for incoming single-hop requests.\n\
-                   \# socks_proxy: localhost:9050\n\n\
+                   \# You may need a separate instance of SOCKS proxy for incoming single-hop requests.\n"
+                <> (optDisabled socksProxy <> "socks_proxy: " <> maybe "localhost:9050" (safeDecodeUtf8 . strEncode) socksProxy)
+                <> "\n\n\
                    \# `socks_mode` can be 'onion' for SOCKS proxy to be used for .onion destination hosts only (default)\n\
                    \# or 'always' to be used for all destination hosts (can be used if it is an .onion server).\n\
                    \# socks_mode: onion\n\n\
                    \# Limit number of threads a client can spawn to process proxy commands in parrallel.\n"
-                <> ("# client_concurrency: " <> tshow defaultProxyClientConcurrency <> "\n\n")
-                <> "[INACTIVE_CLIENTS]\n\
+                <> ("# client_concurrency: " <> tshow defaultProxyClientConcurrency)
+                <> "\n\n\
+                   \[INACTIVE_CLIENTS]\n\
                    \# TTL and interval to check inactive clients\n\
-                   \disconnect: off\n"
-                <> ("# ttl: " <> tshow (ttl defaultInactiveClientExpiration) <> "\n")
-                <> ("# check_interval: " <> tshow (checkInterval defaultInactiveClientExpiration) <> "\n")
+                   \disconnect: on\n"
+                <> ("ttl: " <> tshow (ttl defaultInactiveClientExpiration) <> "\n")
+                <> ("check_interval: " <> tshow (checkInterval defaultInactiveClientExpiration))
                 <> "\n\n\
                    \[WEB]\n\
                    \# Set path to generate static mini-site for server information and qr codes/links\n"
                 <> ("static_path: " <> T.pack (fromMaybe defaultStaticPath webStaticPath) <> "\n\n")
                 <> "# Run an embedded server on this port\n\
                    \# Onion sites can use any port and register it in the hidden service config.\n\
-                   \# Running on a port 80 may require setting process capabilities.\n"
-                <> ((if disableWeb then "# " else "") <> "http: 8000\n\n")
-                <> "# You can run an embedded TLS web server too if you provide port and cert and key files.\n\
-                   \# Not required for running relay on onion address.\n\
-                   \# https: 443\n"
-                <> ("# cert: " <> T.pack httpsCertFile <> "\n")
-                <> ("# key: " <> T.pack httpsKeyFile <> "\n")
+                   \# Running on a port 80 may require setting process capabilities.\n\
+                   \# http: 8000\n\n\
+                   \# You can run an embedded TLS web server too if you provide port and cert and key files.\n\
+                   \# Not required for running relay on onion address.\n"
+                <> (webDisabled <> "https: 443\n")
+                <> (webDisabled <> "cert: " <> T.pack httpsCertFile <> "\n")
+                <> (webDisabled <> "key: " <> T.pack httpsKeyFile <> "\n")
+              where
+                webDisabled = if disableWeb then "# " else ""
     runServer ini = do
       hSetBuffering stdout LineBuffering
       hSetBuffering stderr LineBuffering
       fp <- checkSavedFingerprint cfgPath defaultX509Config
       let host = either (const "<hostnames>") T.unpack $ lookupValue "TRANSPORT" "host" ini
           port = T.unpack $ strictIni "TRANSPORT" "port" ini
-          cfg@ServerConfig {information, transports, storeLogFile, newQueueBasicAuth, messageExpiration, inactiveClientExpiration} = serverConfig
+          cfg@ServerConfig {information, storeLogFile, msgStoreType, newQueueBasicAuth, messageExpiration, inactiveClientExpiration} = serverConfig
           sourceCode' = (\ServerPublicInfo {sourceCode} -> sourceCode) <$> information
           srv = ProtoServerWithAuth (SMPServer [THDomainName host] (if port == "5223" then "" else port) (C.KeyHash fp)) newQueueBasicAuth
       printServiceInfo serverVersion srv
       printSourceCode sourceCode'
       printServerConfig transports storeLogFile
+      checkMsgStoreMode msgStoreType
       putStrLn $ case messageExpiration of
         Just ExpirationConfig {ttl} -> "expiring messages after " <> showTTL ttl
         _ -> "not expiring messages"
@@ -244,32 +359,54 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
                 newQueuesAllowed = allowNewQueues cfg,
                 basicAuthEnabled = isJust newQueueBasicAuth
               }
-      runWebServer ini ServerInformation {config, information}
-      runSMPServer cfg
+      case webStaticPath' of
+        Just path | sharedHTTP -> do
+          runWebServer path Nothing ServerInformation {config, information}
+          attachStaticFiles path $ \attachHTTP -> do
+            logDebug "Allocated web server resources"
+            runSMPServer cfg (Just attachHTTP) `finally` logDebug "Releasing web server resources..."
+        Just path -> do
+          runWebServer path webHttpsParams' ServerInformation {config, information}
+          runSMPServer cfg Nothing
+        Nothing -> do
+          logWarn "No server static path set"
+          runSMPServer cfg Nothing
+      logDebug "Bye"
       where
         enableStoreLog = settingIsOn "STORE_LOG" "enable" ini
         logStats = settingIsOn "STORE_LOG" "log_stats" ini
         c = combine cfgPath . ($ defaultX509Config)
+        restoreMessagesFile path = case iniOnOff "STORE_LOG" "restore_messages" ini of
+          Just True -> Just path
+          Just False -> Nothing
+          -- if the setting is not set, it is enabled when store log is enabled
+          _ -> enableStoreLog $> path
+        transports = iniTransports ini
+        sharedHTTP = any (\(_, _, addHTTP) -> addHTTP) transports
+        iniMsgStoreType = either error id $! readMsgStoreType ini
         serverConfig =
           ServerConfig
-            { transports = iniTransports ini,
+            { transports,
               smpHandshakeTimeout = 120000000,
-              tbqSize = 64,
-              -- serverTbqSize = 1024,
-              msgQueueQuota = 128,
+              tbqSize = 128,
+              msgStoreType = iniMsgStoreType,
+              msgQueueQuota = defaultMsgQueueQuota,
+              maxJournalMsgCount = defaultMaxJournalMsgCount,
+              maxJournalStateLines = defaultMaxJournalStateLines,
               queueIdBytes = 24,
               msgIdBytes = 24, -- must be at least 24 bytes, it is used as 192-bit nonce for XSalsa20
-              caCertificateFile = c caCrtFile,
-              privateKeyFile = c serverKeyFile,
-              certificateFile = c serverCrtFile,
+              smpCredentials =
+                ServerCredentials
+                  { caCertificateFile = Just $ c caCrtFile,
+                    privateKeyFile = c serverKeyFile,
+                    certificateFile = c serverCrtFile
+                  },
+              httpCredentials = (\WebHttpsParams {key, cert} -> ServerCredentials {caCertificateFile = Nothing, privateKeyFile = key, certificateFile = cert}) <$> webHttpsParams',
               storeLogFile = enableStoreLog $> storeLogFilePath,
-              storeMsgsFile =
-                let messagesPath = combine logPath "smp-server-messages.log"
-                 in case iniOnOff "STORE_LOG" "restore_messages" ini of
-                      Just True -> Just messagesPath
-                      Just False -> Nothing
-                      -- if the setting is not set, it is enabled when store log is enabled
-                      _ -> enableStoreLog $> messagesPath,
+              storeMsgsFile = case iniMsgStoreType of
+                AMSType SMSMemory -> restoreMessagesFile storeMsgsFilePath
+                AMSType SMSJournal -> Just storeMsgsJournalDir,
+              storeNtfsFile = restoreMessagesFile storeNtfsFilePath,
               -- allow creating new queues by default
               allowNewQueues = fromMaybe True $ iniOnOff "AUTH" "new_queues" ini,
               newQueueBasicAuth = either error id <$!> strDecodeIni "AUTH" "create_password" ini,
@@ -280,6 +417,12 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
                   defaultMessageExpiration
                     { ttl = 86400 * readIniDefault defMsgExpirationDays "STORE_LOG" "expire_messages_days" ini
                     },
+              expireMessagesOnStart = fromMaybe True $ iniOnOff "STORE_LOG" "expire_messages_on_start" ini,
+              idleQueueInterval = defaultIdleQueueInterval,
+              notificationExpiration =
+                defaultNtfExpiration
+                  { ttl = 3600 * readIniDefault defNtfExpirationHours "STORE_LOG" "expire_ntfs_hours" ini
+                  },
               inactiveClientExpiration =
                 settingIsOn "INACTIVE_CLIENTS" "disconnect" ini
                   $> ExpirationConfig
@@ -290,11 +433,14 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
               logStatsStartTime = 0, -- seconds from 00:00 UTC
               serverStatsLogFile = combine logPath "smp-server-stats.daily.log",
               serverStatsBackupFile = logStats $> combine logPath "smp-server-stats.log",
+              prometheusInterval = eitherToMaybe $ read . T.unpack <$> lookupValue "STORE_LOG" "prometheus_interval" ini,
+              prometheusMetricsFile = combine logPath "smp-server-metrics.txt",
+              pendingENDInterval = 15000000, -- 15 seconds
+              ntfDeliveryInterval = 3000000, -- 3 seconds
               smpServerVRange = supportedServerSMPRelayVRange,
               transportConfig =
                 defaultTransportServerConfig
-                  { logTLSErrors = fromMaybe False $ iniOnOff "TRANSPORT" "log_tls_errors" ini,
-                    alpn = Just supportedSMPHandshakes
+                  { logTLSErrors = fromMaybe False $ iniOnOff "TRANSPORT" "log_tls_errors" ini
                   },
               controlPort = eitherToMaybe $ T.unpack <$> lookupValue "TRANSPORT" "control_port" ini,
               smpAgentCfg =
@@ -306,8 +452,8 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
                           networkConfig =
                             defaultNetworkConfig
                               { socksProxy = either error id <$!> strDecodeIni "PROXY" "socks_proxy" ini,
-                                socksMode = either (const SMOnion) textToSocksMode $ lookupValue "PROXY" "socks_mode" ini,
-                                hostMode = either (const HMPublic) textToHostMode $ lookupValue "PROXY" "host_mode" ini,
+                                socksMode = maybe SMOnion (either error id) $! strDecodeIni "PROXY" "socks_mode" ini,
+                                hostMode = either (const HMPublic) (either error id . textToHostMode) $ lookupValue "PROXY" "host_mode" ini,
                                 requiredHostMode = fromMaybe False $ iniOnOff "PROXY" "required_host_mode" ini
                               }
                         },
@@ -318,38 +464,50 @@ smpServerCLI_ generateSite serveStaticFiles cfgPath logPath =
               serverClientConcurrency = readIniDefault defaultProxyClientConcurrency "PROXY" "client_concurrency" ini,
               information = serverPublicInfo ini
             }
-        textToSocksMode :: Text -> SocksMode
-        textToSocksMode = \case
-          "always" -> SMAlways
-          "onion" -> SMOnion
-          s -> error . T.unpack $ "Invalid socks_mode: " <> s
-        textToHostMode :: Text -> HostMode
-        textToHostMode = \case
-          "public" -> HMPublic
-          "onion" -> HMOnionViaSocks
-          s -> error . T.unpack $ "Invalid host_mode: " <> s
         textToOwnServers :: Text -> [ByteString]
         textToOwnServers = map encodeUtf8 . T.words
-
-    runWebServer ini si =
-      case eitherToMaybe $ T.unpack <$> lookupValue "WEB" "static_path" ini of
-        Nothing -> logWarn "No server static path set"
-        Just webStaticPath -> do
+        runWebServer webStaticPath webHttpsParams si = do
           let onionHost =
                 either (const Nothing) (find isOnion) $
                   strDecode @(L.NonEmpty TransportHost) . encodeUtf8 =<< lookupValue "TRANSPORT" "host" ini
               webHttpPort = eitherToMaybe $ read . T.unpack <$> lookupValue "WEB" "http" ini
-              webHttpsParams =
-                eitherToMaybe $ do
-                  port <- read . T.unpack <$> lookupValue "WEB" "https" ini
-                  cert <- T.unpack <$> lookupValue "WEB" "cert" ini
-                  key <- T.unpack <$> lookupValue "WEB" "key" ini
-                  pure WebHttpsParams {port, cert, key}
           generateSite si onionHost webStaticPath
           when (isJust webHttpPort || isJust webHttpsParams) $
             serveStaticFiles EmbeddedWebParams {webStaticPath, webHttpPort, webHttpsParams}
           where
             isOnion = \case THOnionHost _ -> True; _ -> False
+        webHttpsParams' =
+          eitherToMaybe $ do
+            port <- read . T.unpack <$> lookupValue "WEB" "https" ini
+            cert <- T.unpack <$> lookupValue "WEB" "cert" ini
+            key <- T.unpack <$> lookupValue "WEB" "key" ini
+            pure WebHttpsParams {port, cert, key}
+        webStaticPath' = eitherToMaybe $ T.unpack <$> lookupValue "WEB" "static_path" ini
+
+    checkMsgStoreMode :: AMSType -> IO ()
+    checkMsgStoreMode mode = do
+      msgsDirExists <- doesDirectoryExist storeMsgsJournalDir
+      msgsFileExists <- doesFileExist storeMsgsFilePath
+      case mode of
+        _ | msgsFileExists && msgsDirExists -> exitConfigureMsgStorage
+        AMSType SMSJournal
+          | msgsFileExists -> do
+              putStrLn $ "Error: store_messages is `journal` with " <> storeMsgsFilePath <> " file present."
+              putStrLn "Set store_messages to `memory` or use `smp-server journal export` to migrate."
+              exitFailure
+          | not msgsDirExists ->
+              putStrLn $ "store_messages is `journal`, " <> storeMsgsJournalDir <> " directory will be created."
+        AMSType SMSMemory
+          | msgsDirExists -> do
+              putStrLn $ "Error: store_messages is `memory` with " <> storeMsgsJournalDir <> " directory present."
+              putStrLn "Set store_messages to `journal` or use `smp-server journal import` to migrate."
+              exitFailure
+        _ -> pure ()
+
+    exitConfigureMsgStorage = do
+      putStrLn $ "Error: both " <> storeMsgsFilePath <> " file and " <> storeMsgsJournalDir <> " directory are present."
+      putStrLn "Configure memory storage."
+      exitFailure
 
 data EmbeddedWebParams = EmbeddedWebParams
   { webStaticPath :: FilePath,
@@ -373,28 +531,31 @@ getServerSourceCode =
 simplexmqSource :: String
 simplexmqSource = "https://github.com/simplex-chat/simplexmq"
 
-informationIniContent :: Maybe Text -> Text
-informationIniContent sourceCode_ =
+defaultControlPort :: Int
+defaultControlPort = 5224
+
+informationIniContent :: InitOptions -> Text
+informationIniContent InitOptions {sourceCode, serverInfo} =
   "[INFORMATION]\n\
   \# AGPLv3 license requires that you make any source code modifications\n\
   \# available to the end users of the server.\n\
   \# LICENSE: https://github.com/simplex-chat/simplexmq/blob/stable/LICENSE\n\
   \# Include correct source code URI in case the server source code is modified in any way.\n\
   \# If any other information fields are present, source code property also MUST be present.\n\n"
-    <> (maybe "# source_code: URI" ("source_code: " <>) sourceCode_ <> "\n\n")
-    <> "# Declaring all below information is optional, any of these fields can be omitted.\n\
+    <> (optDisabled sourceCode <> "source_code: " <> fromMaybe "URI" sourceCode)
+    <> "\n\n\
+       \# Declaring all below information is optional, any of these fields can be omitted.\n\
        \\n\
        \# Server usage conditions and amendments.\n\
        \# It is recommended to use standard conditions with any amendments in a separate document.\n\
        \# usage_conditions: https://github.com/simplex-chat/simplex-chat/blob/stable/PRIVACY.md\n\
        \# condition_amendments: link\n\
        \\n\
-       \# Server location and operator.\n\
-       \# server_country: ISO-3166 2-letter code\n\
-       \# operator: entity (organization or person name)\n\
-       \# operator_country: ISO-3166 2-letter code\n\
-       \# website:\n\
-       \\n\
+       \# Server location and operator.\n"
+    <> countryStr "server" serverCountry
+    <> enitiyStrs "operator" operator
+    <> (optDisabled website <> "website: " <> fromMaybe "" website)
+    <> "\n\n\
        \# Administrative contacts.\n\
        \# admin_simplex: SimpleX address\n\
        \# admin_email:\n\
@@ -407,9 +568,21 @@ informationIniContent sourceCode_ =
        \# complaints_pgp:\n\
        \# complaints_pgp_fingerprint:\n\
        \\n\
-       \# Hosting provider.\n\
-       \# hosting: entity (organization or person name)\n\
-       \# hosting_country: ISO-3166 2-letter code\n\n"
+       \# Hosting provider.\n"
+    <> enitiyStrs "hosting" hosting
+    <> "\n\
+       \# Hosting type can be `virtual`, `dedicated`, `colocation`, `owned`\n"
+    <> ("hosting_type: " <> maybe "virtual" (decodeLatin1 . strEncode) hostingType <> "\n\n")
+  where
+    ServerPublicInfo {operator, website, hosting, hostingType, serverCountry} = serverInfo
+    countryStr optName country = optDisabled country <> optName <> "_country: " <> fromMaybe "ISO-3166 2-letter code" country <> "\n"
+    enitiyStrs optName entity =
+      optDisabled entity
+        <> optName
+        <> ": "
+        <> maybe "entity (organization or person name)" name entity
+        <> "\n"
+        <> countryStr optName (country =<< entity)
 
 serverPublicInfo :: Ini -> Maybe ServerPublicInfo
 serverPublicInfo ini = serverInfo <$!> infoValue "source_code"
@@ -425,15 +598,14 @@ serverPublicInfo ini = serverInfo <$!> infoValue "source_code"
           website = infoValue "website",
           adminContacts = iniContacts "admin_simplex" "admin_email" "admin_pgp" "admin_pgp_fingerprint",
           complaintsContacts = iniContacts "complaints_simplex" "complaints_email" "complaints_pgp" "complaints_pgp_fingerprint",
-          hosting = iniEntity "hosting" "hosting_country"
+          hosting = iniEntity "hosting" "hosting_country",
+          hostingType = either error id <$!> strDecodeIni "INFORMATION" "hosting_type" ini
         }
     infoValue name = eitherToMaybe $ lookupValue "INFORMATION" name ini
     iniEntity nameField countryField =
       (\name -> Entity {name, country = countryValue countryField})
         <$!> infoValue nameField
-    countryValue field =
-      (\cs -> if T.length cs == 2 && T.all (\c -> isAscii c && isAlpha c) cs then T.map toUpper cs else error $ "Use ISO3166 2-letter code for " <> T.unpack field)
-        <$!> infoValue field
+    countryValue field = (either error id . validCountryValue (T.unpack field) . T.unpack) <$!> infoValue field
     iniContacts simplexField emailField pgpKeyUriField pgpKeyFingerprintField =
       let simplex = either error id . parseAll (connReqUriP' Nothing) . encodeUtf8 <$!> eitherToMaybe (lookupValue "INFORMATION" simplexField ini)
           email = infoValue emailField
@@ -444,6 +616,14 @@ serverPublicInfo ini = serverInfo <$!> infoValue "source_code"
             (Nothing, Nothing, _, Nothing) -> Nothing
             (_, _, pkURI, pkFingerprint) -> Just ServerContactAddress {simplex, email, pgp = PGPKey <$> pkURI <*> pkFingerprint}
 
+optDisabled :: Maybe a -> Text
+optDisabled p = if isNothing p then "# " else ""
+
+validCountryValue :: String -> String -> Either String Text
+validCountryValue field s
+  | length s == 2 && all (\c -> isAscii c && isAlpha c) s = Right $ T.pack $ map toUpper s
+  | otherwise = Left $ "Use ISO3166 2-letter code for " <> field
+
 printSourceCode :: Maybe Text -> IO ()
 printSourceCode = \case
   Just sourceCode -> T.putStrLn $ "Server source code: " <> sourceCode
@@ -456,6 +636,9 @@ data CliCommand
   | OnlineCert CertOptions
   | Start
   | Delete
+  | Journal JournalCmd
+
+data JournalCmd = JCImport | JCExport | JCDelete
 
 data InitOptions = InitOptions
   { enableStoreLog :: Bool,
@@ -464,7 +647,13 @@ data InitOptions = InitOptions
     ip :: HostName,
     fqdn :: Maybe HostName,
     password :: Maybe ServerPassword,
+    controlPort :: Maybe Int,
+    socksProxy :: Maybe SocksProxy,
+    ownDomains :: Maybe (L.NonEmpty TransportHost),
     sourceCode :: Maybe Text,
+    serverInfo :: ServerPublicInfo,
+    operatorCountry :: Maybe Text,
+    hostingCountry :: Maybe Text,
     webStaticPath :: Maybe FilePath,
     disableWeb :: Bool,
     scripted :: Bool
@@ -481,6 +670,7 @@ cliCommandP cfgPath logPath iniFile =
         <> command "cert" (info (OnlineCert <$> certOptionsP) (progDesc $ "Generate new online TLS server credentials (configuration: " <> iniFile <> ")"))
         <> command "start" (info (pure Start) (progDesc $ "Start server (configuration: " <> iniFile <> ")"))
         <> command "delete" (info (pure Delete) (progDesc "Delete configuration and log files"))
+        <> command "journal" (info (Journal <$> journalCmdP) (progDesc "Import/export messages to/from journal storage"))
     )
   where
     initP :: Parser InitOptions
@@ -533,11 +723,45 @@ cliCommandP cfgPath logPath iniFile =
                   <> help "Set password to create new messaging queues"
                   <> value SPRandom
               )
+      controlPort <-
+        flag' (Just defaultControlPort) (long "control-port" <> help ("Enable control port on " <> show defaultControlPort))
+          <|> option strParse (long "control-port" <> help "Enable control port" <> metavar "PORT" <> value Nothing)
+      socksProxy <-
+        flag' (Just defaultSocksProxy) (long "socks-proxy" <> help "Outgoing SOCKS proxy on port 9050")
+          <|> option
+            strParse
+            ( long "socks-proxy"
+                <> metavar "PROXY"
+                <> help "Outgoing SOCKS proxy to forward messages to onion-only servers"
+                <> value Nothing
+            )
+      ownDomains :: Maybe (L.NonEmpty TransportHost) <-
+        option
+          strParse
+          ( long "own-domains"
+              <> metavar "DOMAINS"
+              <> help "Own server domain names (comma-separated)"
+              <> value Nothing
+          )
       sourceCode <-
+        flag' (Just simplexmqSource) (long "source-code" <> help ("Server source code (default: " <> simplexmqSource <> ")"))
+          <|> (optional . strOption) (long "source-code" <> metavar "URI" <> help "Server source code")
+      operator_ <- entityP "operator" "OPERATOR" "Server operator"
+      hosting_ <- entityP "hosting" "HOSTING" "Hosting provider"
+      hostingType <-
+        option
+          strParse
+          ( long "hosting-type"
+              <> metavar "HOSTING_TYPE"
+              <> help "Hosting type: virtual, dedicated, colocation, owned"
+              <> value Nothing
+          )
+      serverCountry <- countryP "server" "SERVER" "Server datacenter"
+      website <-
         (optional . strOption)
-          ( long "source-code"
-              <> help "Server source code will be communicated to the users"
-              <> metavar "URI"
+          ( long "operator-website"
+              <> help "Operator public website"
+              <> metavar "WEBSITE"
           )
       webStaticPath <-
         (optional . strOption)
@@ -564,10 +788,53 @@ cliCommandP cfgPath logPath iniFile =
             ip,
             fqdn,
             password,
-            sourceCode,
+            controlPort,
+            socksProxy,
+            ownDomains,
+            sourceCode = T.pack <$> sourceCode,
+            serverInfo =
+              ServerPublicInfo
+                { sourceCode = T.pack simplexmqSource,
+                  usageConditions = Nothing,
+                  operator = fst operator_,
+                  website,
+                  adminContacts = Nothing,
+                  complaintsContacts = Nothing,
+                  hosting = fst hosting_,
+                  hostingType,
+                  serverCountry
+                },
+            operatorCountry = snd operator_,
+            hostingCountry = snd hosting_,
             webStaticPath,
             disableWeb,
             scripted
           }
+    journalCmdP =
+      hsubparser
+        ( command "import" (info (pure JCImport) (progDesc "Import message log file into a new journal storage"))
+            <> command "export" (info (pure JCExport) (progDesc "Export journal storage to message log file"))
+            <> command "delete" (info (pure JCDelete) (progDesc "Delete journal storage"))
+        )
+
     parseBasicAuth :: ReadM ServerPassword
     parseBasicAuth = eitherReader $ fmap ServerPassword . strDecode . B.pack
+    entityP :: String -> String -> String -> Parser (Maybe Entity, Maybe Text)
+    entityP opt' metavar' help' = do
+      name_ <-
+        (optional . strOption)
+          ( long opt'
+              <> metavar (metavar' <> "_NAME")
+              <> help (help' <> " name")
+          )
+      country <- countryP opt' metavar' help'
+      pure ((\name -> Entity {name, country}) <$> name_, country)
+    countryP :: String -> String -> String -> Parser (Maybe Text)
+    countryP opt' metavar' help' =
+      (optional . option (eitherReader $ validCountryValue opt'))
+        ( long (opt' <> "-country")
+            <> metavar (metavar' <> "_COUNTRY")
+            <> help (help' <> " country")
+        )
+    strParse :: StrEncoding a => ReadM a
+    strParse = eitherReader $ parseAll strP . encodeUtf8 . T.pack

src/Simplex/Messaging/Server/MsgStore/Journal.hs

@@ -0,0 +1,782 @@
+{-# LANGUAGE BangPatterns #-}
+{-# LANGUAGE DataKinds #-}
+{-# LANGUAGE DerivingStrategies #-}
+{-# LANGUAGE DuplicateRecordFields #-}
+{-# LANGUAGE GADTs #-}
+{-# LANGUAGE GeneralizedNewtypeDeriving #-}
+{-# LANGUAGE InstanceSigs #-}
+{-# LANGUAGE LambdaCase #-}
+{-# LANGUAGE MultiWayIf #-}
+{-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE ScopedTypeVariables #-}
+{-# LANGUAGE StandaloneDeriving #-}
+{-# LANGUAGE TypeFamilies #-}
+{-# LANGUAGE TupleSections #-}
+
+module Simplex.Messaging.Server.MsgStore.Journal
+  ( JournalMsgStore (queues, senders, notifiers, random),
+    JournalQueue,
+    JournalMsgQueue (queue, state),
+    JMQueue (queueDirectory, statePath),
+    JournalStoreConfig (..),
+    closeMsgQueue,
+    closeMsgQueueHandles,
+    -- below are exported for tests
+    MsgQueueState (..),
+    JournalState (..),
+    SJournalType (..),
+    msgQueueDirectory,
+    msgQueueStatePath,
+    readWriteQueueState,
+    newMsgQueueState,
+    newJournalId,
+    appendState,
+    queueLogFileName,
+    journalFilePath,
+    logFileExt,
+  )
+where
+
+import Control.Concurrent.STM
+import qualified Control.Exception as E
+import Control.Logger.Simple
+import Control.Monad
+import Control.Monad.Trans.Except
+import qualified Data.Attoparsec.ByteString.Char8 as A
+import Data.ByteString.Char8 (ByteString)
+import qualified Data.ByteString.Char8 as B
+import Data.Functor (($>))
+import Data.Int (Int64)
+import Data.List (intercalate)
+import Data.Maybe (catMaybes, fromMaybe, isNothing)
+import qualified Data.Text as T
+import Data.Time.Clock (getCurrentTime)
+import Data.Time.Clock.System (SystemTime (..), getSystemTime)
+import Data.Time.Format.ISO8601 (iso8601Show)
+import GHC.IO (catchAny)
+import Simplex.Messaging.Agent.Client (getMapLock, withLockMap)
+import Simplex.Messaging.Agent.Lock
+import Simplex.Messaging.Encoding.String
+import Simplex.Messaging.Protocol
+import Simplex.Messaging.Server.MsgStore.Types
+import Simplex.Messaging.Server.QueueStore
+import Simplex.Messaging.Server.QueueStore.STM
+import Simplex.Messaging.TMap (TMap)
+import qualified Simplex.Messaging.TMap as TM
+import Simplex.Messaging.Server.StoreLog
+import Simplex.Messaging.Util (ifM, tshow, ($>>=), (<$$>))
+import System.Directory
+import System.Exit
+import System.FilePath ((</>))
+import System.IO (BufferMode (..), Handle, IOMode (..), SeekMode (..), stdout)
+import qualified System.IO as IO
+import System.Random (StdGen, genByteString, newStdGen)
+
+data JournalMsgStore = JournalMsgStore
+  { config :: JournalStoreConfig,
+    random :: TVar StdGen,
+    queueLocks :: TMap RecipientId Lock,
+    queues :: TMap RecipientId JournalQueue,
+    senders :: TMap SenderId RecipientId,
+    notifiers :: TMap NotifierId RecipientId,
+    storeLog :: TVar (Maybe (StoreLog 'WriteMode))
+  }
+
+data JournalStoreConfig = JournalStoreConfig
+  { storePath :: FilePath,
+    pathParts :: Int,
+    quota :: Int,
+    -- Max number of messages per journal file - ignored in STM store.
+    -- When this limit is reached, the file will be changed.
+    -- This number should be set bigger than queue quota.
+    maxMsgCount :: Int,
+    maxStateLines :: Int,
+    stateTailSize :: Int,
+    -- time in seconds after which the queue will be closed after message expiration
+    idleInterval :: Int64
+  }
+
+data JournalQueue = JournalQueue
+  { queueLock :: Lock,
+    -- To avoid race conditions and errors when restoring queues,
+    -- Nothing is written to TVar when queue is deleted.
+    queueRec :: TVar (Maybe QueueRec),
+    msgQueue_ :: TVar (Maybe JournalMsgQueue),
+    -- system time in seconds since epoch
+    activeAt :: TVar Int64,
+    -- Just True - empty, Just False - non-empty, Nothing - unknown
+    isEmpty :: TVar (Maybe Bool)
+  }
+
+data JMQueue = JMQueue
+  { queueDirectory :: FilePath,
+    statePath :: FilePath
+  }
+
+data JournalMsgQueue = JournalMsgQueue
+  { queue :: JMQueue,
+    state :: TVar MsgQueueState,
+    -- tipMsg contains last message and length incl. newline
+    -- Nothing - unknown, Just Nothing - empty queue.
+    -- It  prevents reading each message twice,
+    -- and reading it after it was just written.
+    tipMsg :: TVar (Maybe (Maybe (Message, Int64))),
+    handles :: TVar (Maybe MsgQueueHandles)
+  }
+
+data MsgQueueState = MsgQueueState
+  { readState :: JournalState 'JTRead,
+    writeState :: JournalState 'JTWrite,
+    canWrite :: Bool,
+    size :: Int
+  }
+  deriving (Show)
+
+data MsgQueueHandles = MsgQueueHandles
+  { stateHandle :: Handle, -- handle to queue state log file, rotates and removes old backups when server is restarted
+    readHandle :: Handle,
+    writeHandle :: Maybe Handle -- optional, used when write file is different from read file
+  }
+
+data JournalState t = JournalState
+  { journalType :: SJournalType t,
+    journalId :: ByteString,
+    msgPos :: Int,
+    msgCount :: Int,
+    bytePos :: Int64,
+    byteCount :: Int64
+  }
+  deriving (Show)
+
+data JournalType = JTRead | JTWrite
+
+data SJournalType (t :: JournalType) where
+  SJTRead :: SJournalType 'JTRead
+  SJTWrite :: SJournalType 'JTWrite
+
+class JournalTypeI t where sJournalType :: SJournalType t
+
+instance JournalTypeI 'JTRead where sJournalType = SJTRead
+
+instance JournalTypeI 'JTWrite where sJournalType = SJTWrite
+
+deriving instance Show (SJournalType t)
+
+newMsgQueueState :: ByteString -> MsgQueueState
+newMsgQueueState journalId =
+  MsgQueueState
+    { writeState = newJournalState journalId,
+      readState = newJournalState journalId,
+      canWrite = True,
+      size = 0
+    }
+
+newJournalState :: JournalTypeI t => ByteString -> JournalState t
+newJournalState journalId = JournalState sJournalType journalId 0 0 0 0
+
+journalFilePath :: FilePath -> ByteString -> FilePath
+journalFilePath dir journalId = dir </> (msgLogFileName <> "." <> B.unpack journalId <> logFileExt)
+
+instance StrEncoding MsgQueueState where
+  strEncode MsgQueueState {writeState, readState, canWrite, size} =
+    B.unwords
+      [ "write=" <> strEncode writeState,
+        "read=" <> strEncode readState,
+        "canWrite=" <> strEncode canWrite,
+        "size=" <> strEncode size
+      ]
+  strP = do
+    writeState <- "write=" *> strP
+    readState <- " read=" *> strP
+    canWrite <- " canWrite=" *> strP
+    size <- " size=" *> strP
+    pure MsgQueueState {writeState, readState, canWrite, size}
+
+instance JournalTypeI t => StrEncoding (JournalState t) where
+  strEncode JournalState {journalId, msgPos, msgCount, bytePos, byteCount} =
+    B.intercalate "," [journalId, e msgPos, e msgCount, e bytePos, e byteCount]
+    where
+      e :: StrEncoding a => a -> ByteString
+      e = strEncode
+  strP = do
+    journalId <- A.takeTill (== ',')
+    JournalState sJournalType journalId <$> i <*> i <*> i <*> i
+    where
+      i :: Integral a => A.Parser a
+      i = A.char ',' *> A.decimal
+
+queueLogFileName :: String
+queueLogFileName = "queue_state"
+
+msgLogFileName :: String
+msgLogFileName = "messages"
+
+logFileExt :: String
+logFileExt = ".log"
+
+newtype StoreIO a = StoreIO {unStoreIO :: IO a}
+  deriving newtype (Functor, Applicative, Monad)
+
+instance STMQueueStore JournalMsgStore where
+  queues' = queues
+  senders' = senders
+  notifiers' = notifiers
+  storeLog' = storeLog
+  mkQueue st qr = do
+    lock <- getMapLock (queueLocks st) $ recipientId qr
+    q <- newTVar $ Just qr
+    mq <- newTVar Nothing
+    activeAt <- newTVar 0
+    isEmpty <- newTVar Nothing
+    pure $ JournalQueue lock q mq activeAt isEmpty
+  msgQueue_' = msgQueue_
+
+instance MsgStoreClass JournalMsgStore where
+  type StoreMonad JournalMsgStore = StoreIO
+  type StoreQueue JournalMsgStore = JournalQueue
+  type MsgQueue JournalMsgStore = JournalMsgQueue
+  type MsgStoreConfig JournalMsgStore = JournalStoreConfig
+
+  newMsgStore :: JournalStoreConfig -> IO JournalMsgStore
+  newMsgStore config = do
+    random <- newTVarIO =<< newStdGen
+    queueLocks <- TM.emptyIO
+    queues <- TM.emptyIO
+    senders <- TM.emptyIO
+    notifiers <- TM.emptyIO
+    storeLog <- newTVarIO Nothing
+    pure JournalMsgStore {config, random, queueLocks, queues, senders, notifiers, storeLog}
+
+  setStoreLog :: JournalMsgStore -> StoreLog 'WriteMode -> IO ()
+  setStoreLog st sl = atomically $ writeTVar (storeLog st) (Just sl)
+
+  closeMsgStore st = do
+    readTVarIO (storeLog st) >>= mapM_ closeStoreLog
+    readTVarIO (queues st) >>= mapM_ closeMsgQueue
+
+  activeMsgQueues = queues
+  {-# INLINE activeMsgQueues #-}
+
+  -- This function is a "foldr" that opens and closes all queues, processes them as defined by action and accumulates the result.
+  -- It is used to export storage to a single file and also to expire messages and validate all queues when server is started.
+  -- TODO this function requires case-sensitive file system, because it uses queue directory as recipient ID.
+  -- It can be made to support case-insensite FS by supporting more than one queue per directory, by getting recipient ID from state file name.
+  withAllMsgQueues :: forall a. Monoid a => Bool -> JournalMsgStore -> (RecipientId -> JournalQueue -> IO a) -> IO a
+  withAllMsgQueues tty ms@JournalMsgStore {config} action = ifM (doesDirectoryExist storePath) processStore (pure mempty)
+    where
+      processStore = do
+        (!count, !res) <- foldQueues 0 processQueue (0, mempty) ("", storePath)
+        putStrLn $ progress count
+        pure res
+      JournalStoreConfig {storePath, pathParts} = config
+      processQueue :: (Int, a) -> (String, FilePath) -> IO (Int, a)
+      processQueue (!i, !r) (queueId, dir) = do
+        when (tty && i `mod` 100 == 0) $ putStr (progress i <> "\r") >> IO.hFlush stdout
+        r' <- case strDecode $ B.pack queueId of
+          Right rId ->
+            getQueue ms SRecipient rId >>= \case
+              Right q -> unStoreIO (getMsgQueue ms rId q) *> action rId q <* closeMsgQueue q
+              Left AUTH -> do
+                logWarn $ "STORE: processQueue, queue " <> T.pack queueId <> " was removed, removing " <> T.pack dir
+                removeQueueDirectory_ dir
+                pure mempty
+              Left e -> do
+                logError $ "STORE: processQueue, error getting queue " <> T.pack queueId <> ", " <> tshow e
+                exitFailure
+          Left e -> do
+            logError $ "STORE: processQueue, message queue directory " <> T.pack dir <> " is invalid, " <> tshow e
+            exitFailure
+        pure (i + 1, r <> r')
+      progress i = "Processed: " <> show i <> " queues"
+      foldQueues depth f acc (queueId, path) = do
+        let f' = if depth == pathParts - 1 then f else foldQueues (depth + 1) f
+        listDirs >>= foldM f' acc
+        where
+          listDirs = fmap catMaybes . mapM queuePath =<< listDirectory path
+          queuePath dir = do
+            let !path' = path </> dir
+                !queueId' = queueId <> dir
+            ifM
+              (doesDirectoryExist path')
+              (pure $ Just (queueId', path'))
+              (Nothing <$ putStrLn ("Error: path " <> path' <> " is not a directory, skipping"))
+
+  logQueueStates :: JournalMsgStore -> IO ()
+  logQueueStates ms = withActiveMsgQueues ms $ \_ -> unStoreIO . logQueueState
+
+  logQueueState :: JournalQueue -> StoreIO ()
+  logQueueState q =
+    StoreIO . void $
+      readTVarIO (msgQueue_ q)
+        $>>= \mq -> readTVarIO (handles mq)
+        $>>= (\hs -> (readTVarIO (state mq) >>= appendState (stateHandle hs)) $> Just ())
+
+  queueRec' = queueRec
+  {-# INLINE queueRec' #-}
+
+  getMsgQueue :: JournalMsgStore -> RecipientId -> JournalQueue -> StoreIO JournalMsgQueue
+  getMsgQueue ms@JournalMsgStore {random} rId JournalQueue {msgQueue_} =
+    StoreIO $ readTVarIO msgQueue_ >>= maybe newQ pure
+    where
+      newQ = do
+        let dir = msgQueueDirectory ms rId
+            statePath = msgQueueStatePath dir $ B.unpack (strEncode rId)
+            queue = JMQueue {queueDirectory = dir, statePath}
+        q <- ifM (doesDirectoryExist dir) (openMsgQueue ms queue) (createQ queue)
+        atomically $ writeTVar msgQueue_ $ Just q
+        pure q
+        where
+          createQ :: JMQueue -> IO JournalMsgQueue
+          createQ queue = do
+            -- folder and files are not created here,
+            -- to avoid file IO for queues without messages during subscription
+            journalId <- newJournalId random
+            mkJournalQueue queue (newMsgQueueState journalId) Nothing
+
+  getPeekMsgQueue :: JournalMsgStore -> RecipientId -> JournalQueue -> StoreIO (Maybe (JournalMsgQueue, Message))
+  getPeekMsgQueue ms rId q@JournalQueue {isEmpty} =
+    StoreIO (readTVarIO isEmpty) >>= \case
+      Just True -> pure Nothing
+      Just False -> peek
+      Nothing -> do
+        -- We only close the queue if we just learnt it's empty.
+        -- This is needed to reduce file descriptors and memory usage
+        -- after the server just started and many clients subscribe.
+        -- In case the queue became non-empty on write and then again empty on read
+        -- we won't be closing it, to avoid frequent open/close on active queues.
+        r <- peek
+        when (isNothing r) $ StoreIO $ closeMsgQueue q
+        pure r
+    where
+      peek = do
+        mq <- getMsgQueue ms rId q
+        (mq,) <$$> tryPeekMsg_ q mq
+
+  -- only runs action if queue is not empty
+  withIdleMsgQueue :: Int64 -> JournalMsgStore -> RecipientId -> JournalQueue -> (JournalMsgQueue -> StoreIO a) -> StoreIO (Maybe a, Int)
+  withIdleMsgQueue now ms@JournalMsgStore {config} rId q action =
+    StoreIO $ readTVarIO (msgQueue_ q) >>= \case
+      Nothing ->
+        E.bracket
+          (unStoreIO $ getPeekMsgQueue ms rId q)
+          (mapM_ $ \_ -> closeMsgQueue q)
+          (maybe (pure (Nothing, 0)) (unStoreIO . run))
+        where
+          run (mq, _) = do
+            r <- action mq
+            sz <- getQueueSize_ mq
+            pure (Just r, sz)
+      Just mq -> do
+        ts <- readTVarIO $ activeAt q
+        r <- if now - ts >= idleInterval config
+          then Just <$> unStoreIO (action mq) `E.finally` closeMsgQueue q
+          else pure Nothing
+        sz <- unStoreIO $ getQueueSize_ mq
+        pure (r, sz)
+
+  deleteQueue :: JournalMsgStore -> RecipientId -> JournalQueue -> IO (Either ErrorType QueueRec)
+  deleteQueue ms rId q =
+    fst <$$> deleteQueue_ ms rId q
+
+  deleteQueueSize :: JournalMsgStore -> RecipientId -> JournalQueue -> IO (Either ErrorType (QueueRec, Int))
+  deleteQueueSize ms rId q =
+    deleteQueue_ ms rId q >>= mapM (traverse getSize)
+    -- traverse operates on the second tuple element
+    where
+      getSize = maybe (pure (-1)) (fmap size . readTVarIO . state)
+
+  getQueueMessages_ :: Bool -> JournalMsgQueue -> StoreIO [Message]
+  getQueueMessages_ drainMsgs q = StoreIO (run [])
+    where
+      run msgs = readTVarIO (handles q) >>= maybe (pure []) (getMsg msgs)
+      getMsg msgs hs = chooseReadJournal q drainMsgs hs >>= maybe (pure msgs) readMsg
+        where
+          readMsg (rs, h) = do
+            (msg, len) <- hGetMsgAt h $ bytePos rs
+            updateReadPos q drainMsgs len hs
+            (msg :) <$> run msgs
+
+  writeMsg :: JournalMsgStore -> RecipientId -> JournalQueue -> Bool -> Message -> ExceptT ErrorType IO (Maybe (Message, Bool))
+  writeMsg ms rId q' logState msg = isolateQueue rId q' "writeMsg" $ do
+    q <- getMsgQueue ms rId q'
+    StoreIO $ (`E.finally` updateActiveAt q') $ do
+      st@MsgQueueState {canWrite, size} <- readTVarIO (state q)
+      let empty = size == 0
+      if canWrite || empty
+        then do
+          atomically $ writeTVar (isEmpty q') (Just False)
+          let canWrt' = quota > size
+          if canWrt'
+            then writeToJournal q st canWrt' msg $> Just (msg, empty)
+            else writeToJournal q st canWrt' msgQuota $> Nothing
+        else pure Nothing
+    where
+      JournalStoreConfig {quota, maxMsgCount} = config ms
+      msgQuota = MessageQuota {msgId = messageId msg, msgTs = messageTs msg}
+      writeToJournal q st@MsgQueueState {writeState, readState = rs, size} canWrt' !msg' = do
+        let msgStr = strEncode msg' `B.snoc` '\n'
+            msgLen = fromIntegral $ B.length msgStr
+        hs <- maybe createQueueDir pure =<< readTVarIO handles
+        (ws, wh) <- case writeHandle hs of
+          Nothing | msgCount writeState >= maxMsgCount -> switchWriteJournal hs
+          wh_ -> pure (writeState, fromMaybe (readHandle hs) wh_)
+        let msgPos' = msgPos ws + 1
+            bytePos' = bytePos ws + msgLen
+            ws' = ws {msgPos = msgPos', msgCount = msgPos', bytePos = bytePos', byteCount = bytePos'}
+            rs' = if journalId ws == journalId rs then rs {msgCount = msgPos', byteCount = bytePos'} else rs
+            !st' = st {writeState = ws', readState = rs', canWrite = canWrt', size = size + 1}
+        hAppend wh (bytePos ws) msgStr
+        updateQueueState q logState hs st' $
+          when (size == 0) $ writeTVar (tipMsg q) $ Just (Just (msg, msgLen))
+        where
+          JournalMsgQueue {queue = JMQueue {queueDirectory, statePath}, handles} = q
+          createQueueDir = do
+            createDirectoryIfMissing True queueDirectory
+            sh <- openFile statePath AppendMode
+            B.hPutStr sh ""
+            rh <- createNewJournal queueDirectory $ journalId rs
+            let hs = MsgQueueHandles {stateHandle = sh, readHandle = rh, writeHandle = Nothing}
+            atomically $ writeTVar handles $ Just hs
+            pure hs
+          switchWriteJournal hs = do
+            journalId <- newJournalId $ random ms
+            wh <- createNewJournal queueDirectory journalId
+            atomically $ writeTVar handles $ Just $ hs {writeHandle = Just wh}
+            pure (newJournalState journalId, wh)
+
+  -- can ONLY be used while restoring messages, not while server running
+  setOverQuota_ :: JournalQueue -> IO ()
+  setOverQuota_ q =
+    readTVarIO (msgQueue_ q)
+      >>= mapM_ (\JournalMsgQueue {state} -> atomically $ modifyTVar' state $ \st -> st {canWrite = False})
+
+  getQueueSize_ :: JournalMsgQueue -> StoreIO Int
+  getQueueSize_ JournalMsgQueue {state} = StoreIO $ size <$> readTVarIO state
+
+  tryPeekMsg_ :: JournalQueue -> JournalMsgQueue -> StoreIO (Maybe Message)
+  tryPeekMsg_ q mq@JournalMsgQueue {tipMsg, handles} =
+    StoreIO $ (readTVarIO handles $>>= chooseReadJournal mq True $>>= peekMsg) >>= setEmpty
+    where
+      peekMsg (rs, h) = readTVarIO tipMsg >>= maybe readMsg (pure . fmap fst)
+        where
+          readMsg = do
+            ml@(msg, _) <- hGetMsgAt h $ bytePos rs
+            atomically $ writeTVar tipMsg $ Just (Just ml)
+            pure $ Just msg
+      setEmpty msg = do
+        atomically $ writeTVar (isEmpty q) (Just $ isNothing msg)
+        pure msg
+
+  tryDeleteMsg_ :: JournalQueue -> JournalMsgQueue -> Bool -> StoreIO ()
+  tryDeleteMsg_ q mq@JournalMsgQueue {tipMsg, handles} logState = StoreIO $ (`E.finally` when logState (updateActiveAt q)) $
+    void $
+      readTVarIO tipMsg -- if there is no cached tipMsg, do nothing
+        $>>= (pure . fmap snd)
+        $>>= \len -> readTVarIO handles
+        $>>= \hs -> updateReadPos mq logState len hs $> Just ()
+
+  isolateQueue :: RecipientId -> JournalQueue -> String -> StoreIO a -> ExceptT ErrorType IO a
+  isolateQueue rId JournalQueue {queueLock} op =
+    tryStore' op rId . withLock' queueLock op . unStoreIO
+
+updateActiveAt :: JournalQueue -> IO ()
+updateActiveAt q = atomically . writeTVar (activeAt q) . systemSeconds =<< getSystemTime
+
+tryStore' :: String -> RecipientId -> IO a -> ExceptT ErrorType IO a
+tryStore' op rId = tryStore op rId . fmap Right
+
+tryStore :: forall a. String -> RecipientId -> IO (Either ErrorType a) -> ExceptT ErrorType IO a
+tryStore op rId a = ExceptT $ E.mask_ $ E.try a >>= either storeErr pure
+  where
+    storeErr :: E.SomeException -> IO (Either ErrorType a)
+    storeErr e =
+      let e' = intercalate ", " [op, B.unpack $ strEncode rId, show e]
+       in logError ("STORE: " <> T.pack e') $> Left (STORE e')
+
+isolateQueueId :: String -> JournalMsgStore -> RecipientId -> IO (Either ErrorType a) -> ExceptT ErrorType IO a
+isolateQueueId op ms rId = tryStore op rId . withLockMap (queueLocks ms) rId op
+
+openMsgQueue :: JournalMsgStore -> JMQueue -> IO JournalMsgQueue
+openMsgQueue ms q@JMQueue {queueDirectory = dir, statePath} = do
+  (st, sh) <- readWriteQueueState ms statePath
+  (st', rh, wh_) <- closeOnException sh $ openJournals ms dir st sh
+  let hs = MsgQueueHandles {stateHandle = sh, readHandle = rh, writeHandle = wh_}
+  mkJournalQueue q st' (Just hs)
+
+mkJournalQueue :: JMQueue -> MsgQueueState -> Maybe MsgQueueHandles -> IO JournalMsgQueue
+mkJournalQueue queue st hs_ = do
+  state <- newTVarIO st
+  tipMsg <- newTVarIO Nothing
+  handles <- newTVarIO hs_
+  -- using the same queue lock which is currently locked,
+  -- to avoid map lookup on queue operations
+  pure JournalMsgQueue {queue, state, tipMsg, handles}
+
+chooseReadJournal :: JournalMsgQueue -> Bool -> MsgQueueHandles -> IO (Maybe (JournalState 'JTRead, Handle))
+chooseReadJournal q log' hs = do
+  st@MsgQueueState {writeState = ws, readState = rs} <- readTVarIO (state q)
+  case writeHandle hs of
+    Just wh | msgPos rs >= msgCount rs && journalId rs /= journalId ws -> do
+      -- switching to write journal
+      atomically $ writeTVar (handles q) $ Just hs {readHandle = wh, writeHandle = Nothing}
+      hClose $ readHandle hs
+      when log' $ removeJournal (queueDirectory $ queue q) rs
+      let !rs' = (newJournalState $ journalId ws) {msgCount = msgCount ws, byteCount = byteCount ws}
+          !st' = st {readState = rs'}
+      updateQueueState q log' hs st' $ pure ()
+      pure $ Just (rs', wh)
+    _ | msgPos rs >= msgCount rs && journalId rs == journalId ws -> pure Nothing
+    _ -> pure $ Just (rs, readHandle hs)
+
+updateQueueState :: JournalMsgQueue -> Bool -> MsgQueueHandles -> MsgQueueState -> STM () -> IO ()
+updateQueueState q log' hs st a = do
+  unless (validQueueState st) $ E.throwIO $ userError $ "updateQueueState invalid state: " <> show st
+  when log' $ appendState (stateHandle hs) st
+  atomically $ writeTVar (state q) st >> a
+
+appendState :: Handle -> MsgQueueState -> IO ()
+appendState h st = E.uninterruptibleMask_ $ B.hPutStr h $ strEncode st `B.snoc` '\n'
+
+updateReadPos :: JournalMsgQueue -> Bool -> Int64 -> MsgQueueHandles -> IO ()
+updateReadPos q log' len hs = do
+  st@MsgQueueState {readState = rs, size} <- readTVarIO (state q)
+  let JournalState {msgPos, bytePos} = rs
+  let msgPos' = msgPos + 1
+      rs' = rs {msgPos = msgPos', bytePos = bytePos + len}
+      st' = st {readState = rs', size = size - 1}
+  updateQueueState q log' hs st' $ writeTVar (tipMsg q) Nothing
+
+msgQueueDirectory :: JournalMsgStore -> RecipientId -> FilePath
+msgQueueDirectory JournalMsgStore {config = JournalStoreConfig {storePath, pathParts}} rId =
+  storePath </> B.unpack (B.intercalate "/" $ splitSegments pathParts $ strEncode rId)
+  where
+    splitSegments _ "" = []
+    splitSegments 1 s = [s]
+    splitSegments n s =
+      let (seg, s') = B.splitAt 2 s
+       in seg : splitSegments (n - 1) s'
+
+msgQueueStatePath :: FilePath -> String -> FilePath
+msgQueueStatePath dir queueId = dir </> (queueLogFileName <> "." <> queueId <> logFileExt)
+
+createNewJournal :: FilePath -> ByteString -> IO Handle
+createNewJournal dir journalId = do
+  let path = journalFilePath dir journalId -- TODO retry if file exists
+  h <- openFile path ReadWriteMode
+  B.hPutStr h ""
+  pure h
+
+newJournalId :: TVar StdGen -> IO ByteString
+newJournalId g = strEncode <$> atomically (stateTVar g $ genByteString 12)
+
+openJournals :: JournalMsgStore -> FilePath -> MsgQueueState -> Handle -> IO (MsgQueueState, Handle, Maybe Handle)
+openJournals ms dir st@MsgQueueState {readState = rs, writeState = ws} sh = do
+  let rjId = journalId rs
+      wjId = journalId ws
+  openJournal rs >>= \case
+    Left path
+      | rjId == wjId -> do
+          logError $ "STORE: openJournals, no read/write file - creating new file, " <> T.pack path
+          newReadJournal
+      | otherwise -> do
+          let rs' = (newJournalState wjId) {msgCount = msgCount ws, byteCount = byteCount ws}
+              st' = st {readState = rs', size = msgCount ws}
+          openJournal rs' >>= \case
+            Left path' -> do
+              logError $ "STORE: openJournals, no read and write files - creating new file, read: " <> T.pack path <> ", write: " <> T.pack path'
+              newReadJournal
+            Right rh -> do
+              logError $ "STORE: openJournals, no read file - switched to write file, " <> T.pack path
+              closeOnException rh $ fixFileSize rh $ bytePos ws
+              pure (st', rh, Nothing)
+    Right rh
+      | rjId == wjId -> do
+          closeOnException rh $ fixFileSize rh $ bytePos ws
+          pure (st, rh, Nothing)
+      | otherwise -> closeOnException rh $ do
+          fixFileSize rh $ byteCount rs
+          openJournal ws >>= \case
+            Left path -> do
+              let msgs = msgCount rs
+                  bytes = byteCount rs
+                  size' = msgs - msgPos rs
+                  ws' = (newJournalState rjId) {msgPos = msgs, msgCount = msgs, bytePos = bytes, byteCount = bytes}
+                  st' = st {writeState = ws', size = size'} -- we don't amend canWrite to trigger QCONT
+              logError $ "STORE: openJournals, no write file, " <> T.pack path
+              pure (st', rh, Nothing)
+            Right wh -> do
+              closeOnException wh $ fixFileSize wh $ bytePos ws
+              pure (st, rh, Just wh)
+  where
+    newReadJournal = do
+      rjId' <- newJournalId $ random ms
+      rh <- createNewJournal dir rjId'
+      let st' = newMsgQueueState rjId'
+      closeOnException rh $ appendState sh st'
+      pure (st', rh, Nothing)
+    openJournal :: JournalState t -> IO (Either FilePath Handle)
+    openJournal JournalState {journalId} =
+      let path = journalFilePath dir journalId
+       in ifM (doesFileExist path) (Right <$> openFile path ReadWriteMode) (pure $ Left path)
+    -- do that for all append operations
+
+fixFileSize :: Handle -> Int64 -> IO ()
+fixFileSize h pos = do
+  let pos' = fromIntegral pos
+  size <- IO.hFileSize h
+  if
+    | size > pos' -> do
+        name <- IO.hShow h
+        logWarn $ "STORE: fixFileSize, size " <> tshow size <> " > pos " <> tshow pos <> " - truncating, " <> T.pack name
+        IO.hSetFileSize h pos'
+    | size < pos' -> do
+        -- From code logic this can't happen.
+        name <- IO.hShow h
+        E.throwIO $ userError $ "fixFileSize size " <> show size <> " < pos " <> show pos <> " - aborting: " <> name
+    | otherwise -> pure ()
+
+removeJournal :: FilePath -> JournalState t -> IO ()
+removeJournal dir JournalState {journalId} = do
+  let path = journalFilePath dir journalId
+  removeFile path `catchAny` (\e -> logError $ "STORE: removeJournal, " <> T.pack path <> ", " <> tshow e)
+
+-- This function is supposed to be resilient to crashes while updating state files,
+-- and also resilient to crashes during its execution.
+readWriteQueueState :: JournalMsgStore -> FilePath -> IO (MsgQueueState, Handle)
+readWriteQueueState JournalMsgStore {random, config} statePath =
+  ifM
+    (doesFileExist tempBackup)
+    (renameFile tempBackup statePath >> readQueueState)
+    (ifM (doesFileExist statePath) readQueueState writeNewQueueState)
+  where
+    tempBackup = statePath <> ".bak"
+    readQueueState = do
+      ls <- B.lines <$> readFileTail
+      case ls of
+        [] -> writeNewQueueState
+        _ -> do
+          r@(st, _) <- useLastLine (length ls) True ls
+          unless (validQueueState st) $ E.throwIO $ userError $ "readWriteQueueState inconsistent state: " <> show st
+          pure r
+    writeNewQueueState = do
+      logWarn $ "STORE: readWriteQueueState, empty queue state - initialized, " <> T.pack statePath
+      st <- newMsgQueueState <$> newJournalId random
+      writeQueueState st
+    useLastLine len isLastLine ls = case strDecode $ last ls of
+      Right st
+        | len > maxStateLines config || not isLastLine ->
+            backupWriteQueueState st
+        | otherwise -> do
+            -- when state file has fewer than maxStateLines, we don't compact it
+            sh <- openFile statePath AppendMode
+            pure (st, sh)
+      Left e -- if the last line failed to parse
+        | isLastLine -> case init ls of -- or use the previous line
+            [] -> do
+              logWarn $ "STORE: readWriteQueueState, invalid 1-line queue state - initialized, " <> T.pack statePath
+              st <- newMsgQueueState <$> newJournalId random
+              backupWriteQueueState st
+            ls' -> do
+              logWarn $ "STORE: readWriteQueueState, invalid last line in queue state - using the previous line, " <> T.pack statePath
+              useLastLine len False ls'
+        | otherwise -> E.throwIO $ userError $ "readWriteQueueState invalid state " <> statePath <> ": " <> show e
+    backupWriteQueueState st = do
+      -- State backup is made in two steps to mitigate the crash during the backup.
+      -- Temporary backup file will be used when it is present.
+      renameFile statePath tempBackup -- 1) temp backup
+      r <- writeQueueState st -- 2) save state
+      ts <- getCurrentTime
+      renameFile tempBackup (statePath <> "." <> iso8601Show ts <> ".bak") -- 3) timed backup
+      pure r
+    writeQueueState st = do
+      sh <- openFile statePath AppendMode
+      closeOnException sh $ appendState sh st
+      pure (st, sh)
+    readFileTail =
+      IO.withFile statePath ReadMode $ \h -> do
+        size <- IO.hFileSize h
+        let sz = stateTailSize config
+            sz' = fromIntegral sz
+        if size > sz'
+          then IO.hSeek h AbsoluteSeek (size - sz') >> B.hGet h sz
+          else B.hGet h (fromIntegral size)
+
+validQueueState :: MsgQueueState -> Bool
+validQueueState MsgQueueState {readState = rs, writeState = ws, size}
+  | journalId rs == journalId ws =
+      alwaysValid
+        && msgPos rs <= msgPos ws
+        && msgCount rs == msgCount ws
+        && bytePos rs <= bytePos ws
+        && byteCount rs == byteCount ws
+        && size == msgCount rs - msgPos rs
+  | otherwise =
+      alwaysValid
+        && size == msgCount ws + msgCount rs - msgPos rs
+  where
+    alwaysValid =
+      msgPos rs <= msgCount rs
+        && bytePos rs <= byteCount rs
+        && msgPos ws == msgCount ws
+        && bytePos ws == byteCount ws
+
+deleteQueue_ :: JournalMsgStore -> RecipientId -> JournalQueue -> IO (Either ErrorType (QueueRec, Maybe JournalMsgQueue))
+deleteQueue_ ms rId q =
+  runExceptT $ isolateQueueId "deleteQueue_" ms rId $
+    deleteQueue' ms rId q >>= mapM remove
+  where
+    remove r@(_, mq_) = do
+      mapM_ closeMsgQueueHandles mq_
+      removeQueueDirectory ms rId
+      pure r
+
+closeMsgQueue :: JournalQueue -> IO ()
+closeMsgQueue JournalQueue {msgQueue_} = atomically (swapTVar msgQueue_ Nothing) >>= mapM_ closeMsgQueueHandles
+
+closeMsgQueueHandles :: JournalMsgQueue -> IO ()
+closeMsgQueueHandles q = readTVarIO (handles q) >>= mapM_ closeHandles
+  where
+    closeHandles (MsgQueueHandles sh rh wh_) = do
+      hClose sh
+      hClose rh
+      mapM_ hClose wh_
+
+removeQueueDirectory :: JournalMsgStore -> RecipientId -> IO ()
+removeQueueDirectory st = removeQueueDirectory_ . msgQueueDirectory st
+
+removeQueueDirectory_ :: FilePath -> IO ()
+removeQueueDirectory_ dir =
+  removePathForcibly dir `catchAny` \e ->
+    logError $ "STORE: removeQueueDirectory, " <> T.pack dir <> ", " <> tshow e
+
+hAppend :: Handle -> Int64 -> ByteString -> IO ()
+hAppend h pos s = do
+  fixFileSize h pos
+  IO.hSeek h SeekFromEnd 0
+  B.hPutStr h s
+
+hGetMsgAt :: Handle -> Int64 -> IO (Message, Int64)
+hGetMsgAt h pos = do
+  IO.hSeek h AbsoluteSeek $ fromIntegral pos
+  s <- B.hGetLine h
+  case strDecode s of
+    Right !msg ->
+      let !len = fromIntegral (B.length s) + 1
+       in pure (msg, len)
+    Left e -> E.throwIO $ userError $ "hGetMsgAt invalid message: " <> e
+
+openFile :: FilePath -> IOMode -> IO Handle
+openFile f mode = do
+  h <- IO.openFile f mode
+  IO.hSetBuffering h LineBuffering
+  pure h
+
+hClose :: Handle -> IO ()
+hClose h =
+  IO.hClose h `catchAny` \e -> do
+    name <- IO.hShow h
+    logError $ "STORE: hClose, " <> T.pack name <> ", " <> tshow e
+
+closeOnException :: Handle -> IO a -> IO a
+closeOnException h a = a `E.onException` hClose h

src/Simplex/Messaging/Server/MsgStore/STM.hs

@@ -1,133 +1,175 @@
-{-# LANGUAGE BangPatterns #-}
 {-# LANGUAGE ConstraintKinds #-}
+{-# LANGUAGE DataKinds #-}
+{-# LANGUAGE DuplicateRecordFields #-}
 {-# LANGUAGE FlexibleContexts #-}
 {-# LANGUAGE FlexibleInstances #-}
+{-# LANGUAGE InstanceSigs #-}
 {-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE MultiParamTypeClasses #-}
 {-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE TypeFamilies #-}
 {-# LANGUAGE TupleSections #-}
 
 module Simplex.Messaging.Server.MsgStore.STM
-  ( STMMsgStore,
-    MsgQueue (..),
-    newMsgStore,
-    getMsgQueue,
-    delMsgQueue,
-    delMsgQueueSize,
-    flushMsgQueue,
-    snapshotMsgQueue,
-    writeMsg,
-    tryPeekMsg,
-    peekMsg,
-    tryDelMsg,
-    tryDelPeekMsg,
-    deleteExpiredMsgs,
+  ( STMMsgStore (..),
+    STMStoreConfig (..),
   )
 where
 
-import Control.Concurrent.STM.TQueue (flushTQueue)
-import qualified Data.ByteString.Char8 as B
+import Control.Concurrent.STM
+import Control.Monad.IO.Class
+import Control.Monad.Trans.Except
 import Data.Functor (($>))
 import Data.Int (Int64)
-import Data.Time.Clock.System (SystemTime (systemSeconds))
-import Simplex.Messaging.Protocol (Message (..), MsgId, RecipientId)
+import Simplex.Messaging.Protocol
+import Simplex.Messaging.Server.MsgStore.Types
+import Simplex.Messaging.Server.QueueStore
+import Simplex.Messaging.Server.QueueStore.STM
+import Simplex.Messaging.Server.StoreLog
 import Simplex.Messaging.TMap (TMap)
 import qualified Simplex.Messaging.TMap as TM
-import UnliftIO.STM
+import Simplex.Messaging.Util ((<$$>), ($>>=))
+import System.IO (IOMode (..))
 
-data MsgQueue = MsgQueue
+data STMMsgStore = STMMsgStore
+  { storeConfig :: STMStoreConfig,
+    queues :: TMap RecipientId STMQueue,
+    senders :: TMap SenderId RecipientId,
+    notifiers :: TMap NotifierId RecipientId,
+    storeLog :: TVar (Maybe (StoreLog 'WriteMode))
+  }
+
+data STMQueue = STMQueue
+  { -- To avoid race conditions and errors when restoring queues,
+    -- Nothing is written to TVar when queue is deleted.
+    queueRec :: TVar (Maybe QueueRec),
+    msgQueue_ :: TVar (Maybe STMMsgQueue)
+  }
+
+data STMMsgQueue = STMMsgQueue
   { msgQueue :: TQueue Message,
-    quota :: Int,
     canWrite :: TVar Bool,
     size :: TVar Int
   }
 
-type STMMsgStore = TMap RecipientId MsgQueue
+data STMStoreConfig = STMStoreConfig
+  { storePath :: Maybe FilePath,
+    quota :: Int
+  }
 
-newMsgStore :: STM STMMsgStore
-newMsgStore = TM.empty
+instance STMQueueStore STMMsgStore where
+  queues' = queues
+  senders' = senders
+  notifiers' = notifiers
+  storeLog' = storeLog
+  mkQueue _ qr = STMQueue <$> newTVar (Just qr) <*> newTVar Nothing
+  msgQueue_' = msgQueue_
 
-getMsgQueue :: STMMsgStore -> RecipientId -> Int -> STM MsgQueue
-getMsgQueue st rId quota = maybe newQ pure =<< TM.lookup rId st
-  where
-    newQ = do
-      msgQueue <- newTQueue
-      canWrite <- newTVar True
-      size <- newTVar 0
-      let q = MsgQueue {msgQueue, quota, canWrite, size}
-      TM.insert rId q st
-      pure q
+instance MsgStoreClass STMMsgStore where
+  type StoreMonad STMMsgStore = STM
+  type StoreQueue STMMsgStore = STMQueue
+  type MsgQueue STMMsgStore = STMMsgQueue
+  type MsgStoreConfig STMMsgStore = STMStoreConfig
 
-delMsgQueue :: STMMsgStore -> RecipientId -> STM ()
-delMsgQueue st rId = TM.delete rId st
+  newMsgStore :: STMStoreConfig -> IO STMMsgStore
+  newMsgStore storeConfig = do
+    queues <- TM.emptyIO
+    senders <- TM.emptyIO
+    notifiers <- TM.emptyIO
+    storeLog <- newTVarIO Nothing
+    pure STMMsgStore {storeConfig, queues, senders, notifiers, storeLog}
 
-delMsgQueueSize :: STMMsgStore -> RecipientId -> STM Int
-delMsgQueueSize st rId = TM.lookupDelete rId st >>= maybe (pure 0) (\MsgQueue {size} -> readTVar size)
+  setStoreLog :: STMMsgStore -> StoreLog 'WriteMode -> IO ()
+  setStoreLog st sl = atomically $ writeTVar (storeLog st) (Just sl)
 
-flushMsgQueue :: STMMsgStore -> RecipientId -> STM [Message]
-flushMsgQueue st rId = TM.lookupDelete rId st >>= maybe (pure []) (flushTQueue . msgQueue)
+  closeMsgStore st = readTVarIO (storeLog st) >>= mapM_ closeStoreLog
 
-snapshotMsgQueue :: STMMsgStore -> RecipientId -> STM [Message]
-snapshotMsgQueue st rId = TM.lookup rId st >>= maybe (pure []) (snapshotTQueue . msgQueue)
-  where
-    snapshotTQueue q = do
-      msgs <- flushTQueue q
-      mapM_ (writeTQueue q) msgs
-      pure msgs
+  activeMsgQueues = queues
+  {-# INLINE activeMsgQueues #-}
 
-writeMsg :: MsgQueue -> Message -> STM (Maybe Message)
-writeMsg MsgQueue {msgQueue = q, quota, canWrite, size} !msg = do
-  canWrt <- readTVar canWrite
-  empty <- isEmptyTQueue q
-  if canWrt || empty
-    then do
-      canWrt' <- (quota >) <$> readTVar size
-      writeTVar canWrite $! canWrt'
-      modifyTVar' size (+ 1)
-      if canWrt'
-        then writeTQueue q msg $> Just msg
-        else (writeTQueue q $! msgQuota) $> Nothing
-    else pure Nothing
-  where
-    msgQuota = MessageQuota {msgId = msgId msg, msgTs = msgTs msg}
+  withAllMsgQueues _ = withActiveMsgQueues
+  {-# INLINE withAllMsgQueues #-}
 
-tryPeekMsg :: MsgQueue -> STM (Maybe Message)
-tryPeekMsg = tryPeekTQueue . msgQueue
-{-# INLINE tryPeekMsg #-}
+  logQueueStates _ = pure ()
 
-peekMsg :: MsgQueue -> STM Message
-peekMsg = peekTQueue . msgQueue
-{-# INLINE peekMsg #-}
+  logQueueState _ = pure ()
 
-tryDelMsg :: MsgQueue -> MsgId -> STM (Maybe Message)
-tryDelMsg mq msgId' =
-  tryPeekMsg mq >>= \case
-    msg_@(Just msg)
-      | msgId msg == msgId' || B.null msgId' -> tryDeleteMsg mq >> pure msg_
-      | otherwise -> pure Nothing
-    _ -> pure Nothing
+  queueRec' = queueRec
+  {-# INLINE queueRec' #-}
 
--- atomic delete (== read) last and peek next message if available
-tryDelPeekMsg :: MsgQueue -> MsgId -> STM (Maybe Message, Maybe Message)
-tryDelPeekMsg mq msgId' =
-  tryPeekMsg mq >>= \case
-    msg_@(Just msg)
-      | msgId msg == msgId' || B.null msgId' -> (msg_,) <$> (tryDeleteMsg mq >> tryPeekMsg mq)
-      | otherwise -> pure (Nothing, msg_)
-    _ -> pure (Nothing, Nothing)
+  getMsgQueue :: STMMsgStore -> RecipientId -> STMQueue -> STM STMMsgQueue
+  getMsgQueue _ _ STMQueue {msgQueue_} = readTVar msgQueue_ >>= maybe newQ pure
+    where
+      newQ = do
+        msgQueue <- newTQueue
+        canWrite <- newTVar True
+        size <- newTVar 0
+        let q = STMMsgQueue {msgQueue, canWrite, size}
+        writeTVar msgQueue_ (Just q)
+        pure q
 
-deleteExpiredMsgs :: MsgQueue -> Int64 -> STM Int
-deleteExpiredMsgs mq old = loop 0
-  where
-    loop dc =
-      tryPeekMsg mq >>= \case
-        Just Message {msgTs}
-          | systemSeconds msgTs < old ->
-              tryDeleteMsg mq >> loop (dc + 1)
-        _ -> pure dc
+  getPeekMsgQueue :: STMMsgStore -> RecipientId -> STMQueue -> STM (Maybe (STMMsgQueue, Message))
+  getPeekMsgQueue _ _ q@STMQueue {msgQueue_} = readTVar msgQueue_ $>>= \mq -> (mq,) <$$> tryPeekMsg_ q mq
 
-tryDeleteMsg :: MsgQueue -> STM ()
-tryDeleteMsg MsgQueue {msgQueue = q, size} =
-  tryReadTQueue q >>= \case
-    Just _ -> modifyTVar' size (subtract 1)
-    _ -> pure ()
+  -- does not create queue if it does not exist, does not delete it if it does (can't just close in-memory queue)
+  withIdleMsgQueue :: Int64 -> STMMsgStore -> RecipientId -> STMQueue -> (STMMsgQueue -> STM a) -> STM (Maybe a, Int)
+  withIdleMsgQueue _ _ _ STMQueue {msgQueue_} action = readTVar msgQueue_ >>= \case
+    Just q -> do
+      r <- action q
+      sz <- getQueueSize_ q
+      pure (Just r, sz)
+    Nothing -> pure (Nothing, 0)
+
+  deleteQueue :: STMMsgStore -> RecipientId -> STMQueue -> IO (Either ErrorType QueueRec)
+  deleteQueue ms rId q = fst <$$> deleteQueue' ms rId q
+
+  deleteQueueSize :: STMMsgStore -> RecipientId -> STMQueue -> IO (Either ErrorType (QueueRec, Int))
+  deleteQueueSize ms rId q = deleteQueue' ms rId q >>= mapM (traverse getSize)
+    -- traverse operates on the second tuple element
+    where
+      getSize = maybe (pure 0) (\STMMsgQueue {size} -> readTVarIO size)
+
+  getQueueMessages_ :: Bool -> STMMsgQueue -> STM [Message]
+  getQueueMessages_ drainMsgs = (if drainMsgs then flushTQueue else snapshotTQueue) . msgQueue
+    where
+      snapshotTQueue q = do
+        msgs <- flushTQueue q
+        mapM_ (writeTQueue q) msgs
+        pure msgs
+
+  writeMsg :: STMMsgStore -> RecipientId -> STMQueue -> Bool -> Message -> ExceptT ErrorType IO (Maybe (Message, Bool))
+  writeMsg ms rId q' _logState msg = liftIO $ atomically $ do
+    STMMsgQueue {msgQueue = q, canWrite, size} <- getMsgQueue ms rId q'
+    canWrt <- readTVar canWrite
+    empty <- isEmptyTQueue q
+    if canWrt || empty
+      then do
+        canWrt' <- (quota >) <$> readTVar size
+        writeTVar canWrite $! canWrt'
+        modifyTVar' size (+ 1)
+        if canWrt'
+          then (writeTQueue q $! msg) $> Just (msg, empty)
+          else (writeTQueue q $! msgQuota) $> Nothing
+      else pure Nothing
+    where
+      STMMsgStore {storeConfig = STMStoreConfig {quota}} = ms
+      msgQuota = MessageQuota {msgId = messageId msg, msgTs = messageTs msg}
+
+  setOverQuota_ :: STMQueue -> IO ()
+  setOverQuota_ q = readTVarIO (msgQueue_ q) >>= mapM_ (\mq -> atomically $ writeTVar (canWrite mq) False)
+
+  getQueueSize_ :: STMMsgQueue -> STM Int
+  getQueueSize_ STMMsgQueue {size} = readTVar size
+
+  tryPeekMsg_ :: STMQueue -> STMMsgQueue -> STM (Maybe Message)
+  tryPeekMsg_ _ = tryPeekTQueue . msgQueue
+  {-# INLINE tryPeekMsg_ #-}
+
+  tryDeleteMsg_ :: STMQueue -> STMMsgQueue -> Bool -> STM ()
+  tryDeleteMsg_ _ STMMsgQueue {msgQueue = q, size} _logState =
+    tryReadTQueue q >>= \case
+      Just _ -> modifyTVar' size (subtract 1)
+      _ -> pure ()
+
+  isolateQueue :: RecipientId -> STMQueue -> String -> STM a -> ExceptT ErrorType IO a
+  isolateQueue _ _ _ = liftIO . atomically

src/Simplex/Messaging/Server/MsgStore/Types.hs

@@ -0,0 +1,141 @@
+{-# LANGUAGE BangPatterns #-}
+{-# LANGUAGE DataKinds #-}
+{-# LANGUAGE DuplicateRecordFields #-}
+{-# LANGUAGE FlexibleContexts #-}
+{-# LANGUAGE GADTs #-}
+{-# LANGUAGE LambdaCase #-}
+{-# LANGUAGE MultiWayIf #-}
+{-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE TupleSections #-}
+{-# LANGUAGE TypeFamilyDependencies #-}
+{-# OPTIONS_GHC -Wno-unrecognised-pragmas #-}
+
+{-# HLINT ignore "Redundant multi-way if" #-}
+
+module Simplex.Messaging.Server.MsgStore.Types where
+
+import Control.Concurrent.STM
+import Control.Monad (foldM)
+import Control.Monad.Trans.Except
+import Data.Functor (($>))
+import Data.Int (Int64)
+import Data.Kind
+import qualified Data.Map.Strict as M
+import Data.Time.Clock.System (SystemTime (systemSeconds))
+import Simplex.Messaging.Protocol
+import Simplex.Messaging.Server.QueueStore
+import Simplex.Messaging.Server.StoreLog.Types
+import Simplex.Messaging.TMap (TMap)
+import Simplex.Messaging.Util ((<$$>))
+import System.IO (IOMode (..))
+
+class MsgStoreClass s => STMQueueStore s where
+  queues' :: s -> TMap RecipientId (StoreQueue s)
+  senders' :: s -> TMap SenderId RecipientId
+  notifiers' :: s -> TMap NotifierId RecipientId
+  storeLog' :: s -> TVar (Maybe (StoreLog 'WriteMode))
+  mkQueue :: s -> QueueRec -> STM (StoreQueue s)
+  msgQueue_' :: StoreQueue s -> TVar (Maybe (MsgQueue s))
+
+class Monad (StoreMonad s) => MsgStoreClass s where
+  type StoreMonad s = (m :: Type -> Type) | m -> s
+  type MsgStoreConfig s = c | c -> s
+  type StoreQueue s = q | q -> s
+  type MsgQueue s = q | q -> s
+  newMsgStore :: MsgStoreConfig s -> IO s
+  setStoreLog :: s -> StoreLog 'WriteMode -> IO ()
+  closeMsgStore :: s -> IO ()
+  activeMsgQueues :: s -> TMap RecipientId (StoreQueue s)
+  withAllMsgQueues :: Monoid a => Bool -> s -> (RecipientId -> StoreQueue s -> IO a) -> IO a
+  logQueueStates :: s -> IO ()
+  logQueueState :: StoreQueue s -> StoreMonad s ()
+  queueRec' :: StoreQueue s -> TVar (Maybe QueueRec)
+  getPeekMsgQueue :: s -> RecipientId -> StoreQueue s -> StoreMonad s (Maybe (MsgQueue s, Message))
+  getMsgQueue :: s -> RecipientId -> StoreQueue s -> StoreMonad s (MsgQueue s)
+
+  -- the journal queue will be closed after action if it was initially closed or idle longer than interval in config
+  withIdleMsgQueue :: Int64 -> s -> RecipientId -> StoreQueue s -> (MsgQueue s -> StoreMonad s a) -> StoreMonad s (Maybe a, Int)
+  deleteQueue :: s -> RecipientId -> StoreQueue s -> IO (Either ErrorType QueueRec)
+  deleteQueueSize :: s -> RecipientId -> StoreQueue s -> IO (Either ErrorType (QueueRec, Int))
+  getQueueMessages_ :: Bool -> MsgQueue s -> StoreMonad s [Message]
+  writeMsg :: s -> RecipientId -> StoreQueue s -> Bool -> Message -> ExceptT ErrorType IO (Maybe (Message, Bool))
+  setOverQuota_ :: StoreQueue s -> IO () -- can ONLY be used while restoring messages, not while server running
+  getQueueSize_ :: MsgQueue s -> StoreMonad s Int
+  tryPeekMsg_ :: StoreQueue s -> MsgQueue s -> StoreMonad s (Maybe Message)
+  tryDeleteMsg_ :: StoreQueue s -> MsgQueue s -> Bool -> StoreMonad s ()
+  isolateQueue :: RecipientId -> StoreQueue s -> String -> StoreMonad s a -> ExceptT ErrorType IO a
+
+data MSType = MSMemory | MSJournal
+
+data SMSType :: MSType -> Type where
+  SMSMemory :: SMSType 'MSMemory
+  SMSJournal :: SMSType 'MSJournal
+
+data AMSType = forall s. AMSType (SMSType s)
+
+withActiveMsgQueues :: (MsgStoreClass s, Monoid a) => s -> (RecipientId -> StoreQueue s -> IO a) -> IO a
+withActiveMsgQueues st f = readTVarIO (activeMsgQueues st) >>= foldM run mempty . M.assocs
+  where
+    run !acc (k, v) = do
+      r <- f k v
+      pure $! acc <> r
+
+getQueueMessages :: MsgStoreClass s => Bool -> s -> RecipientId -> StoreQueue s -> ExceptT ErrorType IO [Message]
+getQueueMessages drainMsgs st rId q = withPeekMsgQueue st rId q "getQueueSize" $ maybe (pure []) (getQueueMessages_ drainMsgs . fst)
+{-# INLINE getQueueMessages #-}
+
+getQueueSize :: MsgStoreClass s => s -> RecipientId -> StoreQueue s -> ExceptT ErrorType IO Int
+getQueueSize st rId q = withPeekMsgQueue st rId q "getQueueSize" $ maybe (pure 0) (getQueueSize_ . fst)
+{-# INLINE getQueueSize #-}
+
+tryPeekMsg :: MsgStoreClass s => s -> RecipientId -> StoreQueue s -> ExceptT ErrorType IO (Maybe Message)
+tryPeekMsg st rId q = snd <$$> withPeekMsgQueue st rId q "tryPeekMsg" pure
+{-# INLINE tryPeekMsg #-}
+
+tryDelMsg :: MsgStoreClass s => s -> RecipientId -> StoreQueue s -> MsgId -> ExceptT ErrorType IO (Maybe Message)
+tryDelMsg st rId q msgId' =
+  withPeekMsgQueue st rId q "tryDelMsg" $
+    maybe (pure Nothing) $ \(mq, msg) ->
+      if
+        | messageId msg == msgId' ->
+            tryDeleteMsg_ q mq True $> Just msg
+        | otherwise -> pure Nothing
+
+-- atomic delete (== read) last and peek next message if available
+tryDelPeekMsg :: MsgStoreClass s => s -> RecipientId -> StoreQueue s -> MsgId -> ExceptT ErrorType IO (Maybe Message, Maybe Message)
+tryDelPeekMsg st rId q msgId' =
+  withPeekMsgQueue st rId q "tryDelPeekMsg" $
+    maybe (pure (Nothing, Nothing)) $ \(mq, msg) ->
+      if
+        | messageId msg == msgId' -> (Just msg,) <$> (tryDeleteMsg_ q mq True >> tryPeekMsg_ q mq)
+        | otherwise -> pure (Nothing, Just msg)
+
+-- The action is called with Nothing when it is known that the queue is empty
+withPeekMsgQueue :: MsgStoreClass s => s -> RecipientId -> StoreQueue s -> String -> (Maybe (MsgQueue s, Message) -> StoreMonad s a) -> ExceptT ErrorType IO a
+withPeekMsgQueue st rId q op a = isolateQueue rId q op $ getPeekMsgQueue st rId q >>= a
+{-# INLINE withPeekMsgQueue #-}
+
+deleteExpiredMsgs :: MsgStoreClass s => s -> RecipientId -> StoreQueue s -> Int64 -> ExceptT ErrorType IO Int
+deleteExpiredMsgs st rId q old =
+  isolateQueue rId q "deleteExpiredMsgs" $
+    getMsgQueue st rId q >>= deleteExpireMsgs_ old q
+
+-- closed and idle queues will be closed after expiration
+-- returns (expired count, queue size after expiration)
+idleDeleteExpiredMsgs :: MsgStoreClass s => Int64 -> s -> RecipientId -> StoreQueue s -> Int64 -> ExceptT ErrorType IO (Maybe Int, Int)
+idleDeleteExpiredMsgs now st rId q old =
+  isolateQueue rId q "idleDeleteExpiredMsgs" $
+    withIdleMsgQueue now st rId q (deleteExpireMsgs_ old q)
+
+deleteExpireMsgs_ :: MsgStoreClass s => Int64 -> StoreQueue s -> MsgQueue s -> StoreMonad s Int
+deleteExpireMsgs_ old q mq = do
+  n <- loop 0
+  logQueueState q
+  pure n
+  where
+    loop dc =
+      tryPeekMsg_ q mq >>= \case
+        Just Message {msgTs}
+          | systemSeconds msgTs < old ->
+              tryDeleteMsg_ q mq False >> loop (dc + 1)
+        _ -> pure dc

src/Simplex/Messaging/Server/NtfStore.hs

@@ -0,0 +1,67 @@
+{-# LANGUAGE BangPatterns #-}
+{-# LANGUAGE LambdaCase #-}
+{-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE OverloadedStrings #-}
+
+module Simplex.Messaging.Server.NtfStore where
+
+import Control.Concurrent.STM
+import Control.Monad (foldM)
+import Data.Int (Int64)
+import qualified Data.Map.Strict as M
+import Data.Time.Clock.System (SystemTime (..))
+import qualified Simplex.Messaging.Crypto as C
+import Simplex.Messaging.Encoding.String
+import Simplex.Messaging.Protocol (EncNMsgMeta, MsgId, NotifierId)
+import Simplex.Messaging.TMap (TMap)
+import qualified Simplex.Messaging.TMap as TM
+
+newtype NtfStore = NtfStore (TMap NotifierId (TVar [MsgNtf]))
+
+data MsgNtf = MsgNtf
+  { ntfMsgId :: MsgId,
+    ntfTs :: SystemTime,
+    ntfNonce :: C.CbNonce,
+    ntfEncMeta :: EncNMsgMeta
+  }
+
+storeNtf :: NtfStore -> NotifierId -> MsgNtf -> IO ()
+storeNtf (NtfStore ns) nId ntf = do
+  TM.lookupIO nId ns >>= atomically . maybe newNtfs (`modifyTVar'` (ntf :))
+  -- TODO coalesce messages here once the client is updated to process multiple messages
+  -- for single notification.
+  -- when (isJust prevNtf) $ incStat $ msgNtfReplaced stats
+  where
+    newNtfs = TM.lookup nId ns >>= maybe (TM.insertM nId (newTVar [ntf]) ns) (`modifyTVar'` (ntf :))
+
+deleteNtfs :: NtfStore -> NotifierId -> IO Int
+deleteNtfs (NtfStore ns) nId = atomically (TM.lookupDelete nId ns) >>= maybe (pure 0) (fmap length . readTVarIO)
+
+deleteExpiredNtfs :: NtfStore -> Int64 -> IO Int
+deleteExpiredNtfs (NtfStore ns) old =
+  foldM (\expired -> fmap (expired +) . expireQueue) 0 . M.keys =<< readTVarIO ns
+  where
+    expireQueue nId = TM.lookupIO nId ns >>= maybe (pure 0) expire
+    expire v = readTVarIO v >>= \case
+      [] -> pure 0
+      _ ->
+        atomically $ readTVar v >>= \case
+          [] -> pure 0
+          -- check the last message first, it is the earliest
+          ntfs | systemSeconds (ntfTs $ last $ ntfs) < old -> do
+            let !ntfs' = filter (\MsgNtf {ntfTs = ts} -> systemSeconds ts >= old) ntfs
+            writeTVar v ntfs'
+            pure $! length ntfs - length ntfs'
+          _ -> pure 0
+
+data NtfLogRecord = NLRv1 NotifierId MsgNtf
+
+instance StrEncoding MsgNtf where
+  strEncode MsgNtf {ntfMsgId, ntfTs, ntfNonce, ntfEncMeta} = strEncode (ntfMsgId, ntfTs, ntfNonce, ntfEncMeta)
+  strP = do
+    (ntfMsgId, ntfTs, ntfNonce, ntfEncMeta) <- strP
+    pure MsgNtf {ntfMsgId, ntfTs, ntfNonce, ntfEncMeta}
+
+instance StrEncoding NtfLogRecord where
+  strEncode (NLRv1 nId ntf) = strEncode (Str "v1", nId, ntf)
+  strP = "v1 " *> (NLRv1 <$> strP_ <*> strP)

src/Simplex/Messaging/Server/Prometheus.hs

@@ -0,0 +1,379 @@
+{-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE TypeApplications #-}
+{-# OPTIONS_GHC -fno-warn-unrecognised-pragmas #-}
+
+module Simplex.Messaging.Server.Prometheus where
+
+import Data.Int (Int64)
+import Data.Text (Text)
+import qualified Data.Text as T
+import Data.Time.Clock (UTCTime (..), diffUTCTime)
+import Data.Time.Clock.System (systemEpochDay)
+import Data.Time.Format.ISO8601 (iso8601Show)
+import Network.Socket (ServiceName)
+import Simplex.Messaging.Server.Stats
+import Simplex.Messaging.Transport.Server (SocketStats (..))
+
+data ServerMetrics = ServerMetrics
+  { statsData :: ServerStatsData,
+    activeQueueCounts :: PeriodStatCounts,
+    activeNtfCounts :: PeriodStatCounts,
+    queueCount :: Int,
+    notifierCount :: Int
+  }
+
+data RealTimeMetrics = RealTimeMetrics
+  { socketStats :: [(ServiceName, SocketStats)],
+    threadsCount :: Int,
+    clientsCount :: Int,
+    smpSubsCount :: Int,
+    smpSubClientsCount :: Int,
+    ntfSubsCount :: Int,
+    ntfSubClientsCount :: Int
+  }
+
+{-# FOURMOLU_DISABLE\n#-}
+prometheusMetrics :: ServerMetrics -> RealTimeMetrics -> UTCTime -> Text
+prometheusMetrics sm rtm ts =
+  time <> queues <> subscriptions <> messages <> ntfMessages <> ntfs <> relays <> info
+  where
+    ServerMetrics {statsData, activeQueueCounts = ps, activeNtfCounts = psNtf, queueCount, notifierCount} = sm
+    RealTimeMetrics
+      { socketStats,
+        threadsCount,
+        clientsCount,
+        smpSubsCount,
+        smpSubClientsCount,
+        ntfSubsCount,
+        ntfSubClientsCount
+      } = rtm
+    ServerStatsData
+      { _fromTime,
+        _qCreated,
+        _qSecured,
+        _qDeletedAll,
+        _qDeletedAllB,
+        _qDeletedNew,
+        _qDeletedSecured,
+        _qSub,
+        _qSubAllB,
+        _qSubAuth,
+        _qSubDuplicate,
+        _qSubProhibited,
+        _qSubEnd,
+        _qSubEndB,
+        _ntfCreated,
+        _ntfDeleted,
+        _ntfDeletedB,
+        _ntfSub,
+        _ntfSubB,
+        _ntfSubAuth,
+        _ntfSubDuplicate,
+        _msgSent,
+        _msgSentAuth,
+        _msgSentQuota,
+        _msgSentLarge,
+        _msgRecv,
+        _msgRecvGet,
+        _msgGet,
+        _msgGetNoMsg,
+        _msgGetAuth,
+        _msgGetDuplicate,
+        _msgGetProhibited,
+        _msgExpired,
+        _activeQueues,
+        _msgSentNtf,
+        _msgRecvNtf,
+        _activeQueuesNtf,
+        _msgNtfs,
+        _msgNtfsB,
+        _msgNtfNoSub,
+        _msgNtfLost,
+        _msgNtfExpired,
+        _pRelays,
+        _pRelaysOwn,
+        _pMsgFwds,
+        _pMsgFwdsOwn,
+        _pMsgFwdsRecv,
+        _qCount,
+        _msgCount,
+        _ntfCount
+      } = statsData
+    time =
+      "# Recorded at: " <> T.pack (iso8601Show ts) <> "\n\
+      \# Stats from: " <> T.pack (iso8601Show _fromTime) <> "\n\
+      \\n"
+    queues =
+      "# Queues\n\
+      \# ------\n\
+      \\n\
+      \# HELP simplex_smp_queues_created Created queues\n\
+      \# TYPE simplex_smp_queues_created counter\n\
+      \simplex_smp_queues_created " <> mshow _qCreated <> "\n# qCreated\n\
+      \\n\
+      \# HELP simplex_smp_queues_secured Secured queues\n\
+      \# TYPE simplex_smp_queues_secured counter\n\
+      \simplex_smp_queues_secured " <> mshow _qSecured <> "\n# qSecured\n\
+      \\n\
+      \# HELP simplex_smp_queues_deleted Deleted queues\n\
+      \# TYPE simplex_smp_queues_deleted counter\n\
+      \simplex_smp_queues_deleted{type=\"all\"} " <> mshow _qDeletedAll <> "\n# qDeleted\n\
+      \simplex_smp_queues_deleted{type=\"new\"} " <> mshow _qDeletedNew <> "\n# qDeletedNew\n\
+      \simplex_smp_queues_deleted{type=\"secured\"} " <> mshow _qDeletedSecured <> "\n# qDeletedSecured\n\
+      \\n\
+      \# HELP simplex_smp_queues_deleted_batch Batched requests to delete queues\n\
+      \# TYPE simplex_smp_queues_deleted_batch counter\n\
+      \simplex_smp_queues_deleted_batch " <> mshow _qDeletedAllB <> "\n# qDeletedAllB\n\
+      \\n\
+      \# HELP simplex_smp_queues_total1 Total number of stored queues (first type of count).\n\
+      \# TYPE simplex_smp_queues_total1 gauge\n\
+      \simplex_smp_queues_total1 " <> mshow _qCount <> "\n# qCount\n\
+      \\n\
+      \# HELP simplex_smp_queues_total2 Total number of stored queues (second type of count).\n\
+      \# TYPE simplex_smp_queues_total2 gauge\n\
+      \simplex_smp_queues_total2 " <> mshow queueCount <> "\n# qCount2\n\
+      \\n\
+      \# HELP simplex_smp_queues_daily Daily active queues.\n\
+      \# TYPE simplex_smp_queues_daily gauge\n\
+      \simplex_smp_queues_daily " <> mstr (dayCount ps) <> "\n# dayMsgQueues\n\
+      \\n\
+      \# HELP simplex_smp_queues_weekly Weekly active queues.\n\
+      \# TYPE simplex_smp_queues_weekly gauge\n\
+      \simplex_smp_queues_weekly " <> mstr (weekCount ps) <> "\n# weekMsgQueues\n\
+      \\n\
+      \# HELP simplex_smp_queues_monthly Monthly active queues.\n\
+      \# TYPE simplex_smp_queues_monthly gauge\n\
+      \simplex_smp_queues_monthly " <> mstr (monthCount ps) <> "\n# monthMsgQueues\n\
+      \\n\
+      \# HELP simplex_smp_queues_notify_daily Daily active queues with notifications.\n\
+      \# TYPE simplex_smp_queues_notify_daily gauge\n\
+      \simplex_smp_queues_notify_daily " <> mstr (dayCount psNtf) <> "\n# dayCountNtf\n\
+      \\n\
+      \# HELP simplex_smp_queues_notify_weekly Weekly active queues with notifications.\n\
+      \# TYPE simplex_smp_queues_notify_weekly gauge\n\
+      \simplex_smp_queues_notify_weekly " <> mstr (weekCount psNtf) <> "\n# weekCountNtf\n\
+      \\n\
+      \# HELP simplex_smp_queues_notify_monthly Monthly active queues with notifications.\n\
+      \# TYPE simplex_smp_queues_notify_monthly gauge\n\
+      \simplex_smp_queues_notify_monthly " <> mstr (monthCount psNtf) <> "\n# monthCountNtf\n\
+      \\n"
+    subscriptions =
+      "# Subscriptions\n\
+      \# -------------\n\
+      \\n\
+      \# HELP simplex_smp_subscribtion_successes Successful subscriptions.\n\
+      \# TYPE simplex_smp_subscribtion_successes counter\n\
+      \simplex_smp_subscribtion_successes " <> mshow _qSub <> "\n# qSub\n\
+      \\n\
+      \# HELP simplex_smp_subscribtion_successes_batch Batched successful subscriptions.\n\
+      \# TYPE simplex_smp_subscribtion_successes_batch counter\n\
+      \simplex_smp_subscribtion_successes_batch " <> mshow _qSubAllB <> "\n# qSubAllB\n\
+      \\n\
+      \# HELP simplex_smp_subscribtion_end Ended subscriptions.\n\
+      \# TYPE simplex_smp_subscribtion_end counter\n\
+      \simplex_smp_subscribtion_end " <> mshow _qSubEnd <> "\n# qSubEnd\n\
+      \\n\
+      \# HELP simplex_smp_subscribtion_end_batch Batched ended subscriptions.\n\
+      \# TYPE simplex_smp_subscribtion_end_batch counter\n\
+      \simplex_smp_subscribtion_end_batch " <> mshow _qSubEndB <> "\n# qSubEndB\n\
+      \\n\
+      \# HELP simplex_smp_subscribtion_errors Subscription errors.\n\
+      \# TYPE simplex_smp_subscribtion_errors counter\n\
+      \simplex_smp_subscribtion_errors{type=\"auth\"} " <> mshow _qSubAuth <> "\n# qSubAuth\n\
+      \simplex_smp_subscribtion_errors{type=\"duplicate\"} " <> mshow _qSubDuplicate <> "\n# qSubDuplicate\n\
+      \simplex_smp_subscribtion_errors{type=\"prohibited\"} " <> mshow _qSubProhibited <> "\n# qSubProhibited\n\
+      \\n"
+    messages =
+      "# Messages\n\
+      \# --------\n\
+      \\n\
+      \# HELP simplex_smp_messages_sent Sent messages.\n\
+      \# TYPE simplex_smp_messages_sent counter\n\
+      \simplex_smp_messages_sent " <> mshow _msgSent <> "\n# msgSent\n\
+      \\n\
+      \# HELP simplex_smp_messages_sent_errors Total number of messages errors by type.\n\
+      \# TYPE simplex_smp_messages_sent_errors counter\n\
+      \simplex_smp_messages_sent_errors{type=\"auth\"} " <> mshow _msgSentAuth <> "\n# msgSentAuth\n\
+      \simplex_smp_messages_sent_errors{type=\"quota\"} " <> mshow _msgSentQuota <> "\n# msgSentQuota\n\
+      \simplex_smp_messages_sent_errors{type=\"large\"} " <> mshow _msgSentLarge <> "\n# msgSentLarge\n\
+      \\n\
+      \# HELP simplex_smp_messages_received Received messages.\n\
+      \# TYPE simplex_smp_messages_received counter\n\
+      \simplex_smp_messages_received " <> mshow _msgRecv <> "\n# msgRecv\n\
+      \\n\
+      \# HELP simplex_smp_messages_expired Expired messages.\n\
+      \# TYPE simplex_smp_messages_expired counter\n\
+      \simplex_smp_messages_expired " <> mshow _msgExpired <> "\n# msgExpired\n\
+      \\n\
+      \# HELP simplex_smp_messages_total Total number of messages stored.\n\
+      \# TYPE simplex_smp_messages_total gauge\n\
+      \simplex_smp_messages_total " <> mshow _msgCount <> "\n# msgCount\n\
+      \\n"
+    ntfMessages =
+      "# Notification messages (client)\n\
+      \# ------------------------------\n\
+      \\n\
+      \# HELP simplex_smp_messages_notify_sent Sent messages with notification flag (cleint).\n\
+      \# TYPE simplex_smp_messages_notify_sent counter\n\
+      \simplex_smp_messages_notify_sent " <> mshow _msgSentNtf <> "\n# msgSentNtf\n\
+      \\n\
+      \# HELP simplex_smp_messages_notify_received Received messages with notification flag (client).\n\
+      \# TYPE simplex_smp_messages_notify_received counter\n\
+      \simplex_smp_messages_notify_received " <> mshow _msgRecvNtf <> "\n# msgRecvNtf\n\
+      \\n\
+      \# HELP simplex_smp_messages_notify_get_sent Requests to get messages with notification flag (client).\n\
+      \# TYPE simplex_smp_messages_notify_get_sent counter\n\
+      \simplex_smp_messages_notify_get_sent " <> mshow _msgGet <> "\n# msgGet\n\
+      \\n\
+      \# HELP simplex_smp_messages_notify_get_received Succesfully received get requests messages with notification flag (client).\n\
+      \# TYPE simplex_smp_messages_notify_get_received counter\n\
+      \simplex_smp_messages_notify_get_received " <> mshow _msgRecvGet <> "\n# msgRecvGet\n\
+      \\n\
+      \# HELP simplex_smp_messages_notify_get_errors Error events with messages with notification flag (client). \n\
+      \# TYPE simplex_smp_messages_notify_get_errors counter\n\
+      \simplex_smp_messages_notify_get_errors{type=\"nomsg\"} " <> mshow _msgGetNoMsg <> "\n# msgGetNoMsg\n\
+      \simplex_smp_messages_notify_get_errors{type=\"auth\"} " <> mshow _msgGetAuth <> "\n# msgGetAuth\n\
+      \simplex_smp_messages_notify_get_errors{type=\"duplicate\"} " <> mshow _msgGetDuplicate <> "\n# msgGetDuplicate\n\
+      \simplex_smp_messages_notify_get_errors{type=\"prohibited\"} " <> mshow _msgGetProhibited <> "\n# msgGetProhibited\n\
+      \\n\
+      \# HELP simplex_smp_queues_notify_created Created queues with notification flag (client).\n\
+      \# TYPE simplex_smp_queues_notify_created counter\n\
+      \simplex_smp_queues_notify_created " <> mshow _ntfCreated <> "\n# ntfCreated\n\
+      \\n\
+      \# HELP simplex_smp_queues_notify_deleted Deleted queues with notification flag (client).\n\
+      \# TYPE simplex_smp_queues_notify_deleted counter\n\
+      \simplex_smp_queues_notify_deleted " <> mshow _ntfDeleted <> "\n# ntfDeleted\n\
+      \\n\
+      \# HELP simplex_smp_queues_notify_deleted_batch Deleted batched queues with notification flag (client).\n\
+      \# TYPE simplex_smp_queues_notify_deleted_batch counter\n\
+      \simplex_smp_queues_notify_deleted_batch " <> mshow _ntfDeletedB <> "\n# ntfDeletedB\n\
+      \\n\
+      \# HELP simplex_smp_queues_notify_total1 Total number of stored queues with notification flag (first type of count).\n\
+      \# TYPE simplex_smp_queues_notify_total1 gauge\n\
+      \simplex_smp_queues_notify_total1 " <> mshow _ntfCount <> "\n# ntfCount1\n\
+      \\n\
+      \# HELP simplex_smp_queues_notify_total2 Total number of stored queues with notification flag (second type of count).\n\
+      \# TYPE simplex_smp_queues_notify_total2 gauge\n\
+      \simplex_smp_queues_notify_total2 " <> mshow notifierCount <> "\n# ntfCount2\n\
+      \\n"
+    ntfs =
+      "# Notifications (server)\n\
+      \# ----------------------\n\
+      \\n\
+      \# HELP simplex_smp_messages_ntf_successes Successful events with notification messages (to ntf server). \n\
+      \# TYPE simplex_smp_messages_ntf_successes counter\n\
+      \simplex_smp_messages_ntf_successes " <> mshow _msgNtfs <> "\n# msgNtfs\n\
+      \\n\
+      \# HELP simplex_smp_messages_ntf_successes_batch Successful batched events with notification messages (to ntf server). \n\
+      \# TYPE simplex_smp_messages_ntf_successes_batch counter\n\
+      \simplex_smp_messages_ntf_successes_batch " <> mshow _msgNtfsB <> "\n# msgNtfsB\n\
+      \\n\
+      \# HELP simplex_smp_messages_ntf_errors Error events with notification messages (to ntf server). \n\
+      \# TYPE simplex_smp_messages_ntf_errors counter\n\
+      \simplex_smp_messages_ntf_errors{type=\"nosub\"} " <> mshow _msgNtfNoSub <> "\n# msgNtfNoSub\n\
+      \simplex_smp_messages_ntf_errors{type=\"lost\"} " <> mshow _msgNtfLost <> "\n# msgNtfLost\n\
+      \simplex_smp_messages_ntf_errors{type=\"expired\"} " <> mshow _msgNtfExpired <> "\n# msgNtfExpired\n\
+      \\n\
+      \# HELP simplex_smp_subscription_ntf_requests Subscription requests with notification flag (from ntf server). \n\
+      \# TYPE simplex_smp_subscription_ntf_requests counter\n\
+      \simplex_smp_subscription_ntf_requests " <> mshow _ntfSub <> "\n# ntfSub\n\
+      \\n\
+      \# HELP simplex_smp_subscription_ntf_requests_batch Batched subscription requests with notification flag (from ntf server). \n\
+      \# TYPE simplex_smp_subscription_ntf_requests_batch counter\n\
+      \simplex_smp_subscription_ntf_requests_batch " <> mshow _ntfSubB <> "\n# ntfSubB\n\
+      \\n\
+      \# HELP simplex_smp_subscribtion_ntf_errors Subscription errors with notification flag (from ntf server). \n\
+      \# TYPE simplex_smp_subscribtion_ntf_errors counter\n\
+      \simplex_smp_subscribtion_ntf_errors{type=\"auth\"} " <> mshow _ntfSubAuth <> "\n# ntfSubAuth\n\
+      \simplex_smp_subscribtion_ntf_errors{type=\"duplicate\"} " <> mshow _ntfSubDuplicate <> "\n# ntfSubDuplicate\n\
+      \\n"
+    relays =
+      "# Relays\n\
+      \# ------\n\
+      \\n\
+      \# HELP simplex_smp_relay_sessions_requests Session requests through relay.\n\
+      \# TYPE simplex_smp_relay_sessions_requests counter\n\
+      \simplex_smp_relay_sessions_requests{source=\"all\"} " <> mshow (_pRequests _pRelays) <> "\n# pRelays_pRequests\n\
+      \simplex_smp_relay_sessions_requests{source=\"own\"} " <> mshow (_pRequests _pRelaysOwn) <> "\n# pRelaysOwn_pRequests\n\
+      \\n\
+      \# HELP simplex_smp_relay_sessions_successes Successful session events through relay.\n\
+      \# TYPE simplex_smp_relay_sessions_successes counter\n\
+      \simplex_smp_relay_sessions_successes{source=\"all\"} " <> mshow (_pSuccesses _pRelays) <> "\n# pRelays_pSuccesses\n\
+      \simplex_smp_relay_sessions_successes{source=\"own\"} " <> mshow (_pSuccesses _pRelaysOwn) <> "\n# pRelaysOwn_pSuccesses\n\
+      \\n\
+      \# HELP simplex_smp_relay_sessions_errors Error session events through relay.\n\
+      \# TYPE simplex_smp_relay_sessions_errors counter\n\
+      \simplex_smp_relay_sessions_errors{source=\"all\",type=\"connect\"} " <> mshow (_pErrorsConnect _pRelays) <> "\n# pRelays_pErrorsConnect\n\
+      \simplex_smp_relay_sessions_errors{source=\"all\",type=\"compat\"} " <> mshow (_pErrorsCompat _pRelays) <> "\n# pRelays_pErrorsCompat\n\
+      \simplex_smp_relay_sessions_errors{source=\"all\",type=\"other\"} " <> mshow (_pErrorsOther _pRelays) <> "\n# pRelays_pErrorsOther\n\
+      \simplex_smp_relay_sessions_errors{source=\"own\",type=\"connect\"} " <> mshow (_pErrorsConnect _pRelaysOwn) <> "\n# pRelaysOwn_pErrorsConnect\n\
+      \simplex_smp_relay_sessions_errors{source=\"own\",type=\"compat\"} " <> mshow (_pErrorsCompat _pRelaysOwn) <> "\n# pRelaysOwn_pErrorsCompat\n\
+      \simplex_smp_relay_sessions_errors{source=\"own\",type=\"other\"} " <> mshow (_pErrorsOther _pRelaysOwn) <> "\n# pRelaysOwn_pErrorsOther\n\
+      \\n\
+      \# HELP simplex_smp_relay_messages_requests Message requests sent through relay.\n\
+      \# TYPE simplex_smp_relay_messages_requests counter\n\
+      \simplex_smp_relay_messages_requests{source=\"all\"} " <> mshow (_pRequests _pMsgFwds) <> "\n# pMsgFwds_pRequests\n\
+      \simplex_smp_relay_messages_requests{source=\"own\"} " <> mshow (_pRequests _pMsgFwdsOwn) <> "\n# pMsgFwdsOwn_pRequests\n\
+      \\n\
+      \# HELP simplex_smp_relay_messages_successes Successful messages sent through relay.\n\
+      \# TYPE simplex_smp_relay_messages_successes counter\n\
+      \simplex_smp_relay_messages_successes{source=\"all\"} " <> mshow (_pSuccesses _pMsgFwds) <> "\n# pMsgFwds_pSuccesses\n\
+      \simplex_smp_relay_messages_successes{source=\"own\"} " <> mshow (_pSuccesses _pMsgFwdsOwn) <> "\n# pMsgFwdsOwn_pSuccesses\n\
+      \\n\
+      \# HELP simplex_smp_relay_messages_errors Error events with messages sent through relay.\n\
+      \# TYPE simplex_smp_relay_messages_errors counter\n\
+      \simplex_smp_relay_messages_errors{source=\"all\",type=\"connect\"} " <> mshow (_pErrorsConnect _pMsgFwds) <> "\n# pMsgFwds_pErrorsConnect\n\
+      \simplex_smp_relay_messages_errors{source=\"all\",type=\"compat\"} " <> mshow (_pErrorsCompat _pMsgFwds) <> "\n# pMsgFwds_pErrorsCompat\n\
+      \simplex_smp_relay_messages_errors{source=\"all\",type=\"other\"} " <> mshow (_pErrorsOther _pMsgFwds) <> "\n# pMsgFwds_pErrorsOther\n\
+      \simplex_smp_relay_messages_errors{source=\"own\",type=\"connect\"} " <> mshow (_pErrorsConnect _pMsgFwdsOwn) <> "\n# pMsgFwdsOwn_pErrorsConnect\n\
+      \simplex_smp_relay_messages_errors{source=\"own\",type=\"compat\"} " <> mshow (_pErrorsCompat _pMsgFwdsOwn) <> "\n# pMsgFwdsOwn_pErrorsCompat\n\
+      \simplex_smp_relay_messages_errors{source=\"own\",type=\"other\"} " <> mshow (_pErrorsOther _pMsgFwdsOwn) <> "\n# pMsgFwdsOwn_pErrorsOther\n\
+      \\n\
+      \# HELP simplex_smp_relay_messages_received Relay messages statistics.\n\
+      \# TYPE simplex_smp_relay_messages_received counter\n\
+      \simplex_smp_relay_messages_received " <> mshow _pMsgFwdsRecv <> "\n# pMsgFwdsRecv\n\
+      \\n"
+    info =
+      "# Info\n\
+      \# ----\n\
+      \\n"
+      <> socketsMetric socketsAccepted "simplex_smp_sockets_accepted" "Accepted sockets"
+      <> socketsMetric socketsClosed "simplex_smp_sockets_closed" "Closed sockets"
+      <> socketsMetric socketsActive "simplex_smp_sockets_active" "Active sockets"
+      <> socketsMetric socketsLeaked "simplex_smp_sockets_leaked" "Leaked sockets"
+      <> "# HELP simplex_smp_threads_total Threads\n\
+      \# TYPE simplex_smp_threads_total gauge\n\
+      \simplex_smp_threads_total " <> mshow threadsCount <> "\n\
+      \\n\
+      \# HELP simplex_smp_clients_total Clients\n\
+      \# TYPE simplex_smp_clients_total gauge\n\
+      \simplex_smp_clients_total " <> mshow clientsCount <> "\n\
+      \\n\
+      \# HELP simplex_smp_subscribtion_total Total subscriptions\n\
+      \# TYPE simplex_smp_subscribtion_total gauge\n\
+      \simplex_smp_subscribtion_total " <> mshow smpSubsCount <> "\n# smpSubs\n\
+      \\n\
+      \# HELP simplex_smp_subscribtion_clients_total Subscribed clients, first counting method\n\
+      \# TYPE simplex_smp_subscribtion_clients_total gauge\n\
+      \simplex_smp_subscribtion_clients_total " <> mshow smpSubClientsCount <> "\n# smpSubClients\n\
+      \\n\
+      \# HELP simplex_smp_subscription_ntf_total Total notification subscripbtions (from ntf server)\n\
+      \# TYPE simplex_smp_subscription_ntf_total gauge\n\
+      \simplex_smp_subscription_ntf_total " <> mshow ntfSubsCount <> "\n# ntfSubs\n\
+      \\n\
+      \# HELP simplex_smp_subscription_ntf_clients_total Total subscribed NTF servers, first counting method\n\
+      \# TYPE simplex_smp_subscription_ntf_clients_total gauge\n\
+      \simplex_smp_subscription_ntf_clients_total " <> mshow ntfSubClientsCount <> "\n# ntfSubClients\n"
+    socketsMetric :: (SocketStats -> Int) -> Text -> Text -> Text
+    socketsMetric sel metric descr =
+      "# HELP " <> metric <> " " <> descr <> "\n"
+        <> "# TYPE " <> metric <> " gauge\n"
+        <> T.concat (map (\(port, ss) -> metric <> "{port=\"" <> T.pack port <> "\"} " <> mshow (sel ss) <> "\n") socketStats)
+        <> "\n"
+    mstr a = T.pack a <> " " <> tsEpoch
+    mshow :: Show a => a -> Text
+    mshow = mstr . show
+    tsEpoch = T.pack $ show @Int64 $ floor @Double $ realToFrac (ts `diffUTCTime` epoch) * 1000
+    epoch = UTCTime systemEpochDay 0
+{-# FOURMOLU_ENABLE\n#-}

src/Simplex/Messaging/Server/QueueStore.hs

@@ -1,10 +1,13 @@
 {-# LANGUAGE DataKinds #-}
+{-# LANGUAGE GeneralizedNewtypeDeriving #-}
 {-# LANGUAGE KindSignatures #-}
 {-# LANGUAGE MultiParamTypeClasses #-}
 {-# LANGUAGE NamedFieldPuns #-}
 
 module Simplex.Messaging.Server.QueueStore where
 
+import Data.Int (Int64)
+import Data.Time.Clock.System (SystemTime (..), getSystemTime)
 import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Protocol
 
@@ -16,7 +19,8 @@ data QueueRec = QueueRec
     senderKey :: !(Maybe SndPublicAuthKey),
     sndSecure :: !SenderCanSecure,
     notifier :: !(Maybe NtfCreds),
-    status :: !ServerQueueStatus
+    status :: !ServerQueueStatus,
+    updatedAt :: !(Maybe RoundedSystemTime)
   }
   deriving (Show)
 
@@ -34,3 +38,16 @@ instance StrEncoding NtfCreds where
     pure NtfCreds {notifierId, notifierKey, rcvNtfDhSecret}
 
 data ServerQueueStatus = QueueActive | QueueOff deriving (Eq, Show)
+
+newtype RoundedSystemTime = RoundedSystemTime Int64
+  deriving (Eq, Ord, Show)
+
+instance StrEncoding RoundedSystemTime where
+  strEncode (RoundedSystemTime t) = strEncode t
+  strP = RoundedSystemTime <$> strP
+
+getRoundedSystemTime :: Int64 -> IO RoundedSystemTime
+getRoundedSystemTime prec = (\t -> RoundedSystemTime $ (systemSeconds t `div` prec) * prec) <$> getSystemTime
+
+getSystemDate :: IO RoundedSystemTime
+getSystemDate = getRoundedSystemTime 86400

src/Simplex/Messaging/Server/QueueStore/STM.hs

@@ -1,112 +1,196 @@
 {-# LANGUAGE BangPatterns #-}
 {-# LANGUAGE DataKinds #-}
+{-# LANGUAGE DuplicateRecordFields #-}
 {-# LANGUAGE FlexibleInstances #-}
 {-# LANGUAGE GADTs #-}
 {-# LANGUAGE KindSignatures #-}
 {-# LANGUAGE LambdaCase #-}
 {-# LANGUAGE MultiParamTypeClasses #-}
 {-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE OverloadedStrings #-}
 {-# LANGUAGE RankNTypes #-}
 {-# LANGUAGE ScopedTypeVariables #-}
+{-# LANGUAGE TupleSections #-}
 {-# LANGUAGE UndecidableInstances #-}
 
 module Simplex.Messaging.Server.QueueStore.STM
-  ( QueueStore (..),
-    newQueueStore,
-    addQueue,
+  ( addQueue,
     getQueue,
+    getQueueRec,
     secureQueue,
     addQueueNotifier,
     deleteQueueNotifier,
     suspendQueue,
-    deleteQueue,
+    updateQueueTime,
+    deleteQueue',
+    readQueueStore,
+    withLog',
   )
 where
 
+import qualified Control.Exception as E
+import Control.Logger.Simple
 import Control.Monad
+import Control.Monad.IO.Class
+import Control.Monad.Trans.Except
+import Data.Bitraversable (bimapM)
+import qualified Data.ByteString.Char8 as B
+import qualified Data.ByteString.Lazy.Char8 as LB
 import Data.Functor (($>))
+import qualified Data.Text as T
+import Data.Text.Encoding (decodeLatin1)
+import Simplex.Messaging.Encoding.String
 import Simplex.Messaging.Protocol
+import Simplex.Messaging.Server.MsgStore.Types
 import Simplex.Messaging.Server.QueueStore
-import Simplex.Messaging.TMap (TMap)
+import Simplex.Messaging.Server.StoreLog
 import qualified Simplex.Messaging.TMap as TM
-import Simplex.Messaging.Util (ifM, ($>>=))
+import Simplex.Messaging.Util (ifM, tshow, ($>>=), (<$$))
+import System.IO
 import UnliftIO.STM
 
-data QueueStore = QueueStore
-  { queues :: TMap RecipientId (TVar QueueRec),
-    senders :: TMap SenderId RecipientId,
-    notifiers :: TMap NotifierId RecipientId
-  }
-
-newQueueStore :: STM QueueStore
-newQueueStore = do
-  queues <- TM.empty
-  senders <- TM.empty
-  notifiers <- TM.empty
-  pure QueueStore {queues, senders, notifiers}
-
-addQueue :: QueueStore -> QueueRec -> STM (Either ErrorType ())
-addQueue QueueStore {queues, senders} q@QueueRec {recipientId = rId, senderId = sId} = do
-  ifM hasId (pure $ Left DUPLICATE_) $ do
-    qVar <- newTVar q
-    TM.insert rId qVar queues
-    TM.insert sId rId senders
-    pure $ Right ()
+addQueue :: STMQueueStore s => s -> QueueRec -> IO (Either ErrorType (StoreQueue s))
+addQueue st qr@QueueRec {recipientId = rId, senderId = sId, notifier}=
+  atomically add
+    $>>= \q -> q <$$ withLog "addQueue" st (`logCreateQueue` qr)
   where
-    hasId = (||) <$> TM.member rId queues <*> TM.member sId senders
+    add = ifM hasId (pure $ Left DUPLICATE_) $ do
+      q <- mkQueue st qr -- STMQueue lock <$> (newTVar $! Just qr) <*> newTVar Nothing
+      TM.insert rId q $ queues' st
+      TM.insert sId rId $ senders' st
+      forM_ notifier $ \NtfCreds {notifierId} -> TM.insert notifierId rId $ notifiers' st
+      pure $ Right q
+    hasId = or <$> sequence [TM.member rId $ queues' st, TM.member sId $ senders' st, hasNotifier]
+    hasNotifier = maybe (pure False) (\NtfCreds {notifierId} -> TM.member notifierId (notifiers' st)) notifier
 
-getQueue :: DirectParty p => QueueStore -> SParty p -> QueueId -> STM (Either ErrorType QueueRec)
-getQueue QueueStore {queues, senders, notifiers} party qId =
-  toResult <$> (mapM readTVar =<< getVar)
+getQueue :: (STMQueueStore s, DirectParty p) => s -> SParty p -> QueueId -> IO (Either ErrorType (StoreQueue s))
+getQueue st party qId =
+  maybe (Left AUTH) Right <$> case party of
+    SRecipient -> TM.lookupIO qId $ queues' st
+    SSender -> TM.lookupIO qId (senders' st) $>>= (`TM.lookupIO` queues' st)
+    SNotifier -> TM.lookupIO qId (notifiers' st) $>>= (`TM.lookupIO` queues' st)
+
+getQueueRec :: (STMQueueStore s, DirectParty p) => s -> SParty p -> QueueId -> IO (Either ErrorType (StoreQueue s, QueueRec))
+getQueueRec st party qId =
+  getQueue st party qId
+    $>>= (\q -> maybe (Left AUTH) (Right . (q,)) <$> readTVarIO (queueRec' q))
+
+secureQueue :: STMQueueStore s => s -> StoreQueue s -> SndPublicAuthKey -> IO (Either ErrorType ())
+secureQueue st sq sKey =
+  atomically (readQueueRec qr $>>= secure)
+    $>>= \rId -> withLog "secureQueue" st $ \s -> logSecureQueue s rId sKey
   where
-    getVar = case party of
-      SRecipient -> TM.lookup qId queues
-      SSender -> TM.lookup qId senders $>>= (`TM.lookup` queues)
-      SNotifier -> TM.lookup qId notifiers $>>= (`TM.lookup` queues)
+    qr = queueRec' sq
+    secure q@QueueRec {recipientId = rId} = case senderKey q of
+      Just k -> pure $ if sKey == k then Right rId else Left AUTH
+      Nothing -> do
+        writeTVar qr $ Just q {senderKey = Just sKey}
+        pure $ Right rId
 
-secureQueue :: QueueStore -> RecipientId -> SndPublicAuthKey -> STM (Either ErrorType QueueRec)
-secureQueue QueueStore {queues} rId sKey =
-  withQueue rId queues $ \qVar ->
-    readTVar qVar >>= \q -> case senderKey q of
-      Just k -> pure $ if sKey == k then Just q else Nothing
-      _ ->
-        let !q' = q {senderKey = Just sKey}
-         in writeTVar qVar q' $> Just q'
+addQueueNotifier :: STMQueueStore s => s -> StoreQueue s -> NtfCreds -> IO (Either ErrorType (Maybe NotifierId))
+addQueueNotifier st sq ntfCreds@NtfCreds {notifierId = nId} =
+  atomically (readQueueRec qr $>>= add)
+    $>>= \(rId, nId_) -> nId_ <$$ withLog "addQueueNotifier" st (\s -> logAddNotifier s rId ntfCreds)
+  where
+    qr = queueRec' sq
+    add q@QueueRec {recipientId = rId} = ifM (TM.member nId (notifiers' st)) (pure $ Left DUPLICATE_) $ do
+      nId_ <- forM (notifier q) $ \NtfCreds {notifierId} -> TM.delete notifierId (notifiers' st) $> notifierId
+      let !q' = q {notifier = Just ntfCreds}
+      writeTVar qr $ Just q'
+      TM.insert nId rId $ notifiers' st
+      pure $ Right (rId, nId_)
 
-addQueueNotifier :: QueueStore -> RecipientId -> NtfCreds -> STM (Either ErrorType QueueRec)
-addQueueNotifier QueueStore {queues, notifiers} rId ntfCreds@NtfCreds {notifierId = nId} = do
-  ifM (TM.member nId notifiers) (pure $ Left DUPLICATE_) $
-    withQueue rId queues $ \qVar -> do
-      q <- readTVar qVar
-      forM_ (notifier q) $ (`TM.delete` notifiers) . notifierId
-      writeTVar qVar $! q {notifier = Just ntfCreds}
-      TM.insert nId rId notifiers
-      pure $ Just q
+deleteQueueNotifier :: STMQueueStore s => s -> StoreQueue s -> IO (Either ErrorType (Maybe NotifierId))
+deleteQueueNotifier st sq =
+  atomically (readQueueRec qr >>= mapM delete)
+    $>>= \(rId, nId_) -> nId_ <$$ withLog "deleteQueueNotifier" st (`logDeleteNotifier` rId)
+  where
+    qr = queueRec' sq
+    delete q = fmap (recipientId q,) $ forM (notifier q) $ \NtfCreds {notifierId} -> do
+      TM.delete notifierId $ notifiers' st
+      writeTVar qr $! Just q {notifier = Nothing}
+      pure notifierId
 
-deleteQueueNotifier :: QueueStore -> RecipientId -> STM (Either ErrorType ())
-deleteQueueNotifier QueueStore {queues, notifiers} rId =
-  withQueue rId queues $ \qVar -> do
-    q <- readTVar qVar
-    forM_ (notifier q) $ \NtfCreds {notifierId} -> TM.delete notifierId notifiers
-    writeTVar qVar $! q {notifier = Nothing}
-    pure $ Just ()
+suspendQueue :: STMQueueStore s => s -> StoreQueue s -> IO (Either ErrorType ())
+suspendQueue st sq =
+  atomically (readQueueRec qr >>= mapM suspend)
+    $>>= \rId -> withLog "suspendQueue" st (`logSuspendQueue` rId)
+  where
+    qr = queueRec' sq
+    suspend q = do
+      writeTVar qr $! Just q {status = QueueOff}
+      pure $ recipientId q
 
-suspendQueue :: QueueStore -> RecipientId -> STM (Either ErrorType ())
-suspendQueue QueueStore {queues} rId =
-  withQueue rId queues $ \qVar -> modifyTVar' qVar (\q -> q {status = QueueOff}) $> Just ()
+updateQueueTime :: STMQueueStore s => s -> StoreQueue s -> RoundedSystemTime -> IO (Either ErrorType QueueRec)
+updateQueueTime st sq t = atomically (readQueueRec qr >>= mapM update) $>>= log'
+  where
+    qr = queueRec' sq
+    update q@QueueRec {updatedAt}
+      | updatedAt == Just t = pure (q, False)
+      | otherwise =
+          let !q' = q {updatedAt = Just t}
+           in (writeTVar qr $! Just q') $> (q', True)
+    log' (q, changed)
+      | changed = q <$$ withLog "updateQueueTime" st (\sl -> logUpdateQueueTime sl (recipientId q) t)
+      | otherwise = pure $ Right q
 
-deleteQueue :: QueueStore -> RecipientId -> STM (Either ErrorType QueueRec)
-deleteQueue QueueStore {queues, senders, notifiers} rId = do
-  TM.lookupDelete rId queues >>= \case
-    Just qVar ->
-      readTVar qVar >>= \q -> do
-        TM.delete (senderId q) senders
-        forM_ (notifier q) $ \NtfCreds {notifierId} -> TM.delete notifierId notifiers
-        pure $ Right q
-    _ -> pure $ Left AUTH
+deleteQueue' :: STMQueueStore s => s -> RecipientId -> StoreQueue s -> IO (Either ErrorType (QueueRec, Maybe (MsgQueue s)))
+deleteQueue' st rId sq =
+  atomically (readQueueRec qr >>= mapM delete)
+    $>>= \q -> withLog "deleteQueue" st (`logDeleteQueue` rId)
+    >>= bimapM pure (\_ -> (q,) <$> atomically (swapTVar (msgQueue_' sq) Nothing))
+  where
+    qr = queueRec' sq
+    delete q = do
+      writeTVar qr Nothing
+      TM.delete (senderId q) $ senders' st
+      forM_ (notifier q) $ \NtfCreds {notifierId} -> TM.delete notifierId $ notifiers' st
+      pure q
 
-toResult :: Maybe a -> Either ErrorType a
-toResult = maybe (Left AUTH) Right
+readQueueRec :: TVar (Maybe QueueRec) -> STM (Either ErrorType QueueRec)
+readQueueRec qr = maybe (Left AUTH) Right <$> readTVar qr
+{-# INLINE readQueueRec #-}
 
-withQueue :: RecipientId -> TMap RecipientId (TVar QueueRec) -> (TVar QueueRec -> STM (Maybe a)) -> STM (Either ErrorType a)
-withQueue rId queues f = toResult <$> TM.lookup rId queues $>>= f
+withLog' :: String -> TVar (Maybe (StoreLog 'WriteMode)) -> (StoreLog 'WriteMode -> IO ()) -> IO (Either ErrorType ())
+withLog' name sl action =
+  readTVarIO sl
+    >>= maybe (pure $ Right ()) (E.try . action >=> bimapM logErr pure)
+  where
+    logErr :: E.SomeException -> IO ErrorType
+    logErr e = logError ("STORE: " <> T.pack err) $> STORE err
+      where
+        err = name <> ", withLog, " <> show e
+
+withLog :: STMQueueStore s => String -> s -> (StoreLog 'WriteMode -> IO ()) -> IO (Either ErrorType ())
+withLog name = withLog' name . storeLog'
+
+readQueueStore :: forall s. STMQueueStore s => FilePath -> s -> IO ()
+readQueueStore f st = withFile f ReadMode $ LB.hGetContents >=> mapM_ processLine . LB.lines
+  where
+    processLine :: LB.ByteString -> IO ()
+    processLine s' = either printError procLogRecord (strDecode s)
+      where
+        s = LB.toStrict s'
+        procLogRecord :: StoreLogRecord -> IO ()
+        procLogRecord = \case
+          CreateQueue q -> addQueue st q >>= qError (recipientId q) "CreateQueue"
+          SecureQueue qId sKey -> withQueue qId "SecureQueue" $ \q -> secureQueue st q sKey
+          AddNotifier qId ntfCreds -> withQueue qId "AddNotifier" $ \q -> addQueueNotifier st q ntfCreds
+          SuspendQueue qId -> withQueue qId "SuspendQueue" $ suspendQueue st
+          DeleteQueue qId -> withQueue qId "DeleteQueue" $ deleteQueue st qId
+          DeleteNotifier qId -> withQueue qId "DeleteNotifier" $ deleteQueueNotifier st
+          UpdateTime qId t -> withQueue qId "UpdateTime" $ \q -> updateQueueTime st q t
+        printError :: String -> IO ()
+        printError e = B.putStrLn $ "Error parsing log: " <> B.pack e <> " - " <> s
+        withQueue :: forall a. RecipientId -> T.Text -> (StoreQueue s -> IO (Either ErrorType a)) -> IO ()
+        withQueue qId op a = runExceptT go >>= qError qId op
+          where
+            go = do
+              q <- ExceptT $ getQueue st SRecipient qId
+              liftIO (readTVarIO $ queueRec' q) >>= \case
+                Nothing -> logWarn $ logPfx qId op <> "already deleted"
+                Just _ -> void $ ExceptT $ a q
+        qError qId op = \case
+          Left e -> logError $ logPfx qId op <> tshow e
+          Right _ -> pure ()
+        logPfx qId op = "STORE: " <> op <> ", stored queue " <> decodeLatin1 (strEncode qId) <> ", "