pg-heartbeat

A long-running Python service should be able to say “I’m still alive” without spinning up a full monitoring stack. pg-heartbeat does exactly that: a HeartbeatHandle writes timestamped rows to a Postgres (or any SQLAlchemy-supported) database, auto-populating hostname, version, and uptime on every beat(). You get a queryable history of each service’s liveness without operating Prometheus, StatsD, or anything in between.

It plays well with multiple replicas via an instance_id field, so each container or worker can record its own beats and you can scope latest() / history() queries to a specific instance. A sibling project to pg-task-tracker — same philosophy: lean on the database you already have.