- Capture container ID before/after 'docker compose up -d' and only send SIGHUP when the same container persisted (config-only change). A recreated container already loaded the fresh config; sending HUP immediately after startup kills Prometheus 3.10 with exit code 2. - Add Prometheus /-/ready gate alongside the postgres/journal health check to fail the deploy if monitoring stack never comes up. - Observed 2026-06-09: infra deploy left trails-cool-prometheus-1 Exited (2) for ~2h, causing a Grafana alert storm. |
||
|---|---|---|
| .. | ||
| workflows | ||
| copilot-instructions.md | ||
| dependabot.yml | ||