Watcher
Operations
Monitoring, retries, and scaling for Watcher
Monitoring
- Sentry is required; job fails if the DSN is missing.
- Logs show pipeline stall warnings with timestamps.
Retries and backfills
- Cloud Run Job retry policy handles transient failures (maxRetries=1).
- Safe to rerun at any time; read-only job.
Resource usage
- Cloud Run Job resources: 1 CPU, 512Mi memory, 120s timeout.
- Single-task execution; no sharding.
Last updated on