Relocate the canonical analytics store off the public VPS. The collector becomes a DB-free edge that captures + durably enqueues events to black's redis; a black-side ingest-writer enriches and writes raw_events. When black is unreachable the collector spools to a local appendonly redis that only grows during the outage and drains on recovery. - shared: RawEventEnvelope/NormalizedEvent ingest contract (edge -> writer) - collector: capture-and-enqueue + dual-redis RedisRouter (primary=black, spool=local) + paused-until-healthy drain worker; drop TypeORM/enrichment - processor: IngestService canonical writer (edge receivedAt, ON CONFLICT DO NOTHING), single worker branches ingest-event -> process-event - relocate device/identity/domain/attribution enrichment + entities to writer Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| scripts | ||
| src | ||
| test | ||
| .swcrc | ||
| Dockerfile | ||
| nest-cli.json | ||
| package.json | ||
| SCALING.md | ||
| tsconfig.json | ||
| vitest.config.ts | ||