observability
Health endpoints, structured logging, correlation IDs, monitoring workflows, and alert hooks.
SKILL.md.claude/skills/08_observability/SKILL.md
GitHubObservability (RESTO BOT)
Health endpoints
| Service | Endpoint | Check |
|---|---|---|
| gateway | http://127.0.0.1:8080/healthz |
wget |
| n8n-main | http://127.0.0.1:5678/healthz |
wget |
| n8n-worker | pgrep -f 'n8n worker' |
process check |
| postgres | pg_isready -U n8n -d n8n |
pg_isready |
| redis | redis-cli ping |
redis-cli |
| cms (Strapi) | http://127.0.0.1:1337/_health |
wget |
| admin-dashboard | http://127.0.0.1:80/ |
wget |
| kiosk-app | http://127.0.0.1:80/ |
wget |
Logging configuration
All services use json-file driver with rotation:
- Critical services (n8n, postgres, redis, gateway, cms): max-size=10m, max-file=5
- Frontend (admin-dashboard, kiosk-app): max-size=5m, max-file=3
Log masking
LOG_MASK_PATTERNS=token,password,secret,api_key,authorization,x-api-token,x-webhook-token,bearer
Structured logging schema
Target fields for every log entry:
- timestamp, level, service, correlation_id
- route, status, latency_ms, error_code (where applicable)
Correlation ID propagation
Flow: gateway (X-Request-ID) -> n8n workflows -> DB writes
SLO monitoring (n8n env vars)
SLO_WINDOW_MIN=15(sliding window)SLO_INBOUND_TO_OUTBOX_P95_MS=2000(2s target)SLO_OUTBOX_PENDING_AGE_MAX_SEC=600(10min max pending)SLO_DLQ_RATE_MAX=0.05(5% max failure rate)SLO_DLQ_COUNT_MAX=5(absolute DLQ cap)
Alert hook
ALERT_WEBHOOK_URL env var (can point to Slack, Discord, or monitoring webhook)
Key files
project/docker-compose.hostinger.prod.yml(healthchecks, logging config)project/.env(SLO_*, ALERT_*, LOG_MASK_PATTERNS)project/infra/gateway/nginx.conf(access log format, redaction)project/scripts/smoke_security.sh(health check validation)
Required output
- Health endpoints list + expected outputs
- Logging schema documented
- Smoke tests extended to include health checks
- Incident "what to look at first" checklist