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

Red commit:
https://github.com/Kpa-clawbot/CoreScope/commit/eae179b99b5fd34924547632aa8f8025c405aa53
(CI: pending — opens with this PR)

Finishes #1283. RED test `TestServerSourceHasNoCachedRWCalls` goes from
failing (13 writer call-sites) to GREEN (zero). Per #1287 Option 4
(https://github.com/Kpa-clawbot/CoreScope/issues/1287#issuecomment-4485099992):
ingestor owns the neighbor graph build + persist; server reads the
snapshot.

**Category A — Schema migrations** → new `internal/dbschema` package.
`dbschema.Apply(rw)` runs in `cmd/ingestor` startup (in `OpenStore`).
`dbschema.AssertReady(ro)` runs in `cmd/server/main.go` and
FATAL-LOG-EXITS if any expected column/index/table is missing — the
operator must restart the ingestor first. Covers indexes,
`neighbor_edges`, `observations.resolved_path`,
`observers.{inactive,last_packet_at,iata}`,
`(inactive_)nodes.foreign_advert`, `transmissions.from_pubkey`.

**Category B — Backfill** → ingestor.
`BackfillFromPubkey` and observer-blacklist soft-delete moved to
`cmd/ingestor/maintenance.go`. Server keeps an inert
`fromPubkeyBackfillSnapshot` stub for `/api/healthz` API compatibility.

**Category C — Neighbor-graph persistence (Option 4)** → ingestor
writes, server reads.
- Ingestor (`cmd/ingestor/neighbor_builder.go`): every 60s scans
`observations + transmissions`, extracts edges (originator↔first-hop for
ADVERTs; observer↔last-hop for all), resolves hop prefixes via a
node-table prefix index, upserts into `neighbor_edges`.
- Server (`cmd/server/neighbor_recomputer.go`): every 60s re-reads
`neighbor_edges` and atomic-swaps the resulting `NeighborGraph` into
`s.graph`. Initial load is synchronous on startup. All server-side
incremental edge writers (the two `asyncPersistResolvedPathsAndEdges`
paths in `cmd/server/store.go`) are gone.
- Neighbor-edge daily prune (`PruneNeighborEdges`) moved to ingestor.

**Why Option 4**: clean read/write separation, no startup CPU spike
(server loads existing snapshot instead of rebuilding from history), no
IPC/delta-protocol churn. Staleness budget ~60s — same model as the
analytics recomputers in #1240 / #1248 / #672 axis 2.

**Recomputer interval default for neighbor graph**: 60s
(`NeighborGraphRecomputerDefaultInterval`,
`NeighborEdgesBuilderInterval`).

**Invariants added**:
- `TestServerSourceHasNoCachedRWCalls` (RED commit eae179b9): grep
enforces zero `cachedRW(`, `mode=rw`, or `sql.Open(_journal_mode=WAL…)`
in non-test `cmd/server/` sources.
- `TestServerStartupRequiresMigratedSchema`: server refuses to start
against an unmigrated DB.
- `TestNeighborGraphRecomputerLoadsSnapshot`: post-write snapshot is
picked up on the next refresh.
- `TestNeighborEdgesBuilderUpsertsFromObservations`: end-to-end pipeline
writes the expected edge.

`grep cachedRW cmd/server/*.go | grep -v _test.go` → 0 matches.

Fixes #1287.

---------

Co-authored-by: MeshCore Bot <bot@meshcore.local>
Co-authored-by: Kpa-clawbot <Kpa-clawbot@users.noreply.github.com>
Co-authored-by: corescope-bot <bot@corescope.local>
This commit is contained in:
authored and GitHub committed 2026-05-19 23:53:41 -07:00
1 parent 38b74b6d10
commit 9383201c07
31 files changed
+1514 -2862

No files matched your search

-107
View File
@@ -1,12 +1,8 @@
package main
import (
"database/sql"
"path/filepath"
"testing"
"time"
_ "modernc.org/sqlite"
)
func TestBackfillHoursDefault(t *testing.T) {
@@ -72,106 +68,3 @@ func TestGraphPruneOlderThan(t *testing.T) {
}
}
func TestPruneNeighborEdgesDB(t *testing.T) {
tmpDir := t.TempDir()
dbPath := filepath.Join(tmpDir, "test.db")
db, err := sql.Open("sqlite", "file:"+dbPath+"?_journal_mode=WAL")
if err != nil {
t.Fatal(err)
}
defer db.Close()
_, err = db.Exec(`CREATE TABLE neighbor_edges (
node_a TEXT NOT NULL,
node_b TEXT NOT NULL,
count INTEGER DEFAULT 1,
last_seen TEXT,
PRIMARY KEY (node_a, node_b)
)`)
if err != nil {
t.Fatal(err)
}
now := time.Now().UTC()
old := now.Add(-60 * 24 * time.Hour)
db.Exec("INSERT INTO neighbor_edges (node_a, node_b, count, last_seen) VALUES (?, ?, 5, ?)",
"aaa", "bbb", now.Format(time.RFC3339))
db.Exec("INSERT INTO neighbor_edges (node_a, node_b, count, last_seen) VALUES (?, ?, 3, ?)",
"ccc", "ddd", old.Format(time.RFC3339))
g := NewNeighborGraph()
g.upsertEdge("aaa", "bbb", "bb", "obs1", nil, now)
g.upsertEdge("ccc", "ddd", "dd", "obs1", nil, old)
pruned, err := PruneNeighborEdges(dbPath, g, 30)
if err != nil {
t.Fatal(err)
}
if pruned != 1 {
t.Errorf("PruneNeighborEdges pruned %d DB rows, want 1", pruned)
}
var count int
db.QueryRow("SELECT COUNT(*) FROM neighbor_edges").Scan(&count)
if count != 1 {
t.Errorf("expected 1 row in DB after prune, got %d", count)
}
if len(g.AllEdges()) != 1 {
t.Errorf("expected 1 in-memory edge after prune, got %d", len(g.AllEdges()))
}
}
func TestBackfillRespectsHourWindow(t *testing.T) {
store := &PacketStore{}
now := time.Now().UTC()
oldTime := now.Add(-48 * time.Hour).Format(time.RFC3339Nano)
newTime := now.Add(-30 * time.Minute).Format(time.RFC3339Nano)
store.packets = []*StoreTx{
{
ID: 1,
Hash: "old-hash",
FirstSeen: oldTime,
Observations: []*StoreObs{
{ID: 1, PathJSON: `["abc"]`},
},
},
{
ID: 2,
Hash: "new-hash",
FirstSeen: newTime,
Observations: []*StoreObs{
{ID: 2, PathJSON: `["def"]`},
},
},
}
// With a 1-hour window, only the new tx should be processed.
// backfillResolvedPathsAsync will find no prefix map and finish quickly,
// but we can verify the pending count reflects the window.
go backfillResolvedPathsAsync(store, "", 100, time.Millisecond, 1)
// Wait for completion
for i := 0; i < 100; i++ {
if store.backfillComplete.Load() {
break
}
time.Sleep(10 * time.Millisecond)
}
if !store.backfillComplete.Load() {
t.Fatal("backfill did not complete")
}
// With no prefix map, total should be 0 (early exit) or just the new one
// The function exits early when pm == nil, so backfillTotal stays at 0
// if there were pending items but no pm. Let's verify it didn't process
// the old one by checking total <= 1.
total := store.backfillTotal.Load()
if total > 1 {
t.Errorf("backfill total = %d, want <= 1 (old tx should be excluded by hour window)", total)
}
}