PRD-23: Monitoring & Observability

[PancheI]

Add Pino structured JSON logging, Prometheus metrics with signing/DKG/policy histograms, a pre-built Grafana dashboard, and alert rules.

Acceptance Criteria

• console.log replaced by Pino structured JSON logging in production
• Development mode uses pino-pretty for human-readable output
• GET /metrics returns valid Prometheus exposition format
• guardian_signing_duration_seconds histogram populated after signing
• guardian_signing_requests_total counter with path and status labels
• guardian_active_signing_sessions gauge reflects in-flight sessions
• guardian_dkg_duration_seconds histogram with mode label
• guardian_policy_evaluation_duration_seconds histogram populated
• guardian_policy_violations_total counter per policy type
• guardian_vault_health gauge (1=healthy, 0=sealed/unreachable)
• guardian_http_request_duration_seconds histogram for all requests
• Node.js process metrics collected via collectDefaultMetrics()
• Prometheus scrapes every 15s without errors
• Grafana dashboard loads with all panels
• Alert rules fire correctly (e.g., VaultUnhealthy within 1 minute)
• docker compose up starts Prometheus on :9090 and Grafana on :3001
• No share data in logs, sensitive headers redacted
• Request IDs (reqId) in all log lines for correlation

Tasks

• Install prom-client, nestjs-pino, pino, pino-pretty
• Create MetricsService, MetricsController, MetricsInterceptor
• Configure Pino logger
• Instrument InteractiveSignService, DKGService, AuxInfoPoolService
• Create VaultHealthProbe with periodic check
• Create Prometheus config, alert rules, Grafana provisioning + dashboard JSON
• Update docker-compose.yml with Prometheus + Grafana services
• Unit + integration tests, security audit (no share data in logs)