// the find
CrunchyData/postgres-operator
Production PostgreSQL for Kubernetes, from high availability Postgres clusters to full-scale database-as-a-service.
PGO is a Kubernetes operator that manages production PostgreSQL clusters declaratively — HA failover via Patroni, backups via pgBackRest, connection pooling via pgBouncer, all expressed as CRDs. It's the operator choice for teams running Postgres on Kubernetes who want something backed by a company (Crunchy Data) with genuine Postgres expertise rather than a community experiment.
Patroni-based HA is battle-tested — distributed consensus failover, configurable pod anti-affinity, and cross-cluster standby support are all first-class. pgBackRest integration is thorough: full/incremental/differential backups, delta restores, and multi-repo support (S3, GCS, Azure blob, local) are all wired in. The built-in metrics stack (pgMonitor + Prometheus + Grafana with pre-built dashboards for pgBouncer, pgBackRest, query stats) saves real setup time. The operator handles TLS automatically but lets you bring your own CA if you need to fit into an existing PKI.
The default install pulls from Crunchy Data's developer portal under their own terms of use — not pure Apache 2.0 end-to-end, which matters for teams with strict open-source policies or airgapped environments. Documentation lives entirely at access.crunchydata.com behind a portal login, which is friction when you're debugging at 2am. Major version upgrades use a separate PGUpgrade CRD that still requires careful planning; it's not a one-click affair and the failure modes are not well-documented in the public README. No built-in logical replication management — if you're doing CDC or multi-region writes, you're on your own.