Operations

pino → Alloy → Loki → Grafana four-layer pipeline, label cardinality pitfalls, three deployment modes, when to escalate to Prometheus or OpenTelemetry

Four core dashboards bundled with Zapvol — LogQL queries, read criteria, and troubleshooting paths when something looks wrong

Opt-in stream from dev machines to Grafana Cloud via pino-loki transport — 10-minute setup, no Docker required

End-to-end production deployment using Cloudflare Pages for the frontend and Railway for server, worker, Postgres, and Redis — selection rationale, env var wiring, migration hooks, and pitfall checklist