openclaw-bot 4ea12087f2 fix(mqtt): persistent session + parallel handler (#1337) ...

paho defaults (CleanSession=true, empty random ClientID per reconnect,
Order=true) caused the staging ingestor to receive ~7 msg/h while
mosquitto_sub on the same broker/creds/topics received ~6720/h — a 200x
gap. Every watchdog-driven reconnect (~every 5min) made the broker treat
us as a brand-new session and drop the queued backlog.

buildMQTTOpts now sets:
  - SetClientID("corescope-ingestor-<hostname>-<source-tag>")
    persistent + unique across sources, stable across restarts
  - SetCleanSession(false)
    broker keeps subscription state across reconnects and replays the
    backlog we missed
  - SetKeepAlive(30 * time.Second)
    paho-level half-open detection (was unset; relying on OS keepalive)
  - SetOrderMatters(false)
    handler dispatch is parallel; one slow packet no longer stalls all
    others under burst load

The existing watchdog (#1212/#1216) is untouched. Reconnect throttle
(MaxReconnectInterval=30s) is unchanged — no reconnect storm.

Fixes #1337

2026-05-24 03:00:42 +00:00

..

decrypt

feat: CLI tool to decrypt and export hashtag channel messages (#724)

2026-04-12 22:07:41 -07:00

ingestor

fix(mqtt): persistent session + parallel handler (#1337)

2026-05-24 03:00:42 +00:00

migrate

refactor(db): finish #1283 — Option 4: ingestor owns neighbor-graph + schema migrations; server is read-only (fixes #1287) (#1289)

2026-05-19 23:53:41 -07:00

server

feat: area-based visual node filter — attribute packets by transmitter GPS (#804) (#839)

2026-05-21 14:00:15 -07:00