v0.3.2

Release notes — v0.3.2

First-class PostgreSQL extensions, plus Prometheus scraping for every instance.

Highlights

PostgreSQL extensions

Enable extensions with spec.components.engine.parameters.extensions.<name>.enabled. The provider creates them in the application database through one CloudNativePG Database (<instance>-extensions). That also fixes TimescaleDB instances that never became Ready, because their Database resource was not being synced.

Extension	Toggle	PostgreSQL	Enable later	Disable
pgvector	pgvector	14–18, bundled	Yes, no restart	No
pgAudit	pgaudit	14–18, bundled	Yes, rolling restart	Yes
TimescaleDB OSS	timescaledb	18, image volume	Creation only	No
PostGIS	postgis	18, image volume	Yes, rolling restart	No
pgRouting	pgrouting	18, image volume (needs PostGIS)	Yes, rolling restart	No
pg_ivm	pgivm	18, image volume	Creation only	No
wal2json	wal2json	18, image volume	Yes, rolling restart	Yes
Apache AGE	age	18, image volume	Creation only	No

Image-volume extensions need PostgreSQL 18 and Kubernetes ImageVolume (1.35+, or 1.33/1.34 with the feature gate). Their images come from a CloudNativePG ClusterImageCatalog the chart renders from extensionCatalog (official CloudNativePG trixie builds, plus AGE, pinned by digest). Bump or mirror images with helm upgrade, or set extensionCatalog.create=false and point extensionCatalog.name at a catalog you manage. SQL versions belong in sql-version.extensions.openeverest.io/<major>.<name> annotations. A new image restarts the instances that use it.

TimescaleDB, pg_ivm, and AGE are creation-only. The provider rejects enabling them on a running instance. pgAudit sets pgaudit.log to ddl, role unless you set it in postgresql.parameters. AGE grants the application owner USAGE on ag_catalog at bootstrap. Run SET search_path = ag_catalog, "$user", public before using it. wal2json is a logical decoding plugin, so there is no CREATE EXTENSION; drop replication slots that use it before disabling it.

See examples/instance-pgvector.yaml and examples/instance-timescaledb.yaml.

Prometheus monitoring

With the Prometheus Operator installed, turn on Monitoring → Prometheus to create a PodMonitor (<instance>-metrics) that scrapes the CloudNativePG metrics exporter on port 9187. Scraping uses HTTPS when spec.components.engine.parameters.monitoring.tls.enabled is set. Use podMonitorLabels so your Prometheus podMonitorSelector can select the monitor. Turning the toggle off removes the PodMonitor.

spec:
  topology:
    type: replicaSet
    parameters:
      monitoring:
        prometheus:
          enabled: true
          podMonitorLabels:
            release: kube-prometheus-stack

The PodMonitor is labeled app.kubernetes.io/managed-by=everest and app.kubernetes.io/instance=<instance> for the OpenEverest metrics plugin. The provider rejects spec.components.engine.parameters.monitoring.enablePodMonitor while this toggle is on, so each pod is scraped once.

What's Changed

• Monitoring (PodMonitor) by @spron-in in #85
• First-class PostgreSQL extensions by @spron-in in #86

Full Changelog: v0.3.1...v0.3.2