Files
hyperguild/ingestion/internal/audit/reconcile.go
mathiasandClaude Opus 4.8 77f5e06d6b feat(audit): DegradingSink + durable buffer + loki/ntfy + reconcile (#54)
The classification-aware I5 audit path (§4.4):

- DegradingSink.Reserve: central up → AuditCentral; central down +
  confidential → refuse (no buffer); central down + internal/public +
  buffer writable → AuditBuffered; central down + buffer unwritable →
  floor refuse. Record executes the reserved outcome and, when buffered,
  fires an ntfy alert.
- FileBuffer: durable JSONL buffer that survives process restart; Confirm
  rewrites the file without a record, so a buffered record is cleared ONLY
  after its central write is confirmed.
- LokiCentral: /ready probe + /loki/api/v1/push (full audit entry as the
  structured line). NtfyNotifier: degraded-state alerts; token only in the
  auth header, never logged (regression-tested).
- Reconcile + StartReconcile: replay buffered records to central on
  recovery, confirm-then-clear per record; a failed push keeps the record
  buffered (no loss). SlogSink updated to the two-phase shape (always
  central, never fails) — the default when no loki endpoint is set.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-22 23:54:24 +02:00

66 lines
1.9 KiB
Go

package audit
import (
"context"
"fmt"
"log/slog"
"time"
)
// Reconcile replays locally-buffered audit records to the central sink
// when it is reachable again. A record is removed from the buffer ONLY
// after its central write is confirmed, so a crash mid-reconcile re-plays
// rather than loses. Returns the number of records reconciled.
//
// A no-op (0, nil) when the central sink is still unreachable or the
// buffer is empty.
func Reconcile(ctx context.Context, central Central, buffer Buffer, notifier Notifier) (int, error) {
if err := central.Ready(ctx); err != nil {
return 0, nil // still down; try again next tick
}
pending, err := buffer.Pending()
if err != nil {
return 0, fmt.Errorf("read buffer: %w", err)
}
reconciled := 0
for _, rec := range pending {
if err := central.Push(ctx, rec.Entry); err != nil {
// Central went away mid-drain; stop and keep the rest buffered.
break
}
if err := buffer.Confirm(rec.ID); err != nil {
return reconciled, fmt.Errorf("confirm buffered record %s: %w", rec.ID, err)
}
reconciled++
}
if reconciled > 0 && notifier != nil {
_ = notifier.Notify(ctx, fmt.Sprintf("reconciled %d buffered capture audit record(s) to loki", reconciled))
}
return reconciled, nil
}
// StartReconcile runs Reconcile on a ticker until ctx is cancelled. It is
// the recovery half of the degrade-and-buffer path; pair it with a
// DegradingSink sharing the same buffer + central.
func StartReconcile(ctx context.Context, central Central, buffer Buffer, notifier Notifier, interval time.Duration) {
if interval <= 0 {
interval = time.Minute
}
go func() {
t := time.NewTicker(interval)
defer t.Stop()
for {
select {
case <-ctx.Done():
return
case <-t.C:
if n, err := Reconcile(ctx, central, buffer, notifier); err != nil {
slog.Warn("audit reconcile failed", "err", err)
} else if n > 0 {
slog.Info("audit reconcile", "reconciled", n)
}
}
}
}()
}