BARAAA-70 is now resolved. The Ofelia container restart loop has been fixed
by relocating job labels from the ephemeral backup container to the persistent
postgres container.
Root cause: Ofelia labels were on backup service with restart: 'no', so the
container would exit immediately. Ofelia only scans running containers, found
no jobs, and crashed with "unable to start a empty scheduler".
Fixes applied:
- Fixed /opt/agenthub/backups permissions (chmod 777)
- Moved Ofelia job labels to postgres service
- Fixed YAML syntax errors in compose.lan.yml
Verification: Ofelia now running stably with 0 restarts, backup-daily job
registered with schedule '0 0 3 * * *'.
Next: Monitor backup execution at 3am UTC on 2026-05-03.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Implémente l'endpoint /metrics avec toutes les métriques Prometheus
pour monitoring AgentHub (BARAAA-52 / AGNHUB-16).
Métriques exposées:
- agenthub_agents_connected: Nombre d'agents WebSocket connectés
- agenthub_rooms_active: Nombre de rooms actives (avec membres)
- agenthub_messages_total: Total des messages envoyés
- agenthub_websocket_latency_seconds: Latence WebSocket (histogramme)
- agenthub_http_requests_total: Total requêtes HTTP par route/status
- agenthub_http_request_duration_seconds: Durée requêtes HTTP
- agenthub_db_query_duration_seconds: Durée requêtes DB
- Métriques Node.js par défaut (CPU, mémoire, event loop, etc.)
Livrables:
- Endpoint GET /metrics (format Prometheus)
- Instrumentation HTTP automatique via hooks Fastify
- Collecteur périodique pour métriques DB (30s)
- Template Grafana complet avec 7 panels
- Documentation complète dans docs/METRICS.md
Testé en local - toutes les métriques fonctionnelles.
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Comprehensive summary of J5 development work:
- All deliverables completed
- Deployment blockers documented
- Next steps clearly defined
- Rollback procedures included
Closes development phase of BARAAA-50.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
