checkfleet

Patroni PostgreSQL check

patroni — Patroni PostgreSQL

Cluster topology via the Patroni REST API — a single leader, replica state, replica lag, timeline divergence.

checkfleet check patroni --config checkfleet.yml

Health of a Patroni-managed PostgreSQL cluster from the Patroni REST API (/cluster) — read-only, it never touches PostgreSQL itself.

  • Leader: exactly one expected. Zero leaders → BAD (failover in progress or lost quorum); more than one → WARN (split-brain).
  • Replica state: running/streaming → OK; stopped/start failed/ crashed/restarting → BAD; anything else → WARN.
  • Replica lag: WARN/BAD past lag_warn_bytes/lag_crit_bytes. A lag reported as unknown is OK with a note (not a false alarm).
  • Timeline: a replica on a different timeline than the leader → WARN.

Any Patroni node proxies the same cluster view, so listing one endpoint is enough; extra endpoints add redundancy (an unreachable one is ERROR). The cluster-level leader finding is labelled with the cluster scope; per-node findings are labelled by member name.

See Configuration → checks.patroni.

See the full module reference for every check, or configuration for the config keys.