CloudNativePG provider v0.3.4 adds PgBouncer pooling, high availability, and richer defaults

•

The CloudNativePG provider v0.3.4 is a large day-2 release for PostgreSQL on OpenEverest.

PgBouncer connection pooling

  • A new PgBouncer pooler component pools client connections without a separate proxy to deploy or operate. You declare it alongside the cluster and manage it like every other component.

High availability and recovery

  • The CloudNativePG operator now runs highly available by default, so cluster management survives operator pod failures.
  • Logical replication slots now fail over to replicas, keeping change-data-capture and subscriptions working after a primary switch.
  • New instances can start directly from a VolumeSnapshot, cutting bootstrap and restore times compared to full data restores.
  • Backups now detect a missing barman-cloud plugin and surface a clear alert instead of failing silently.
  • Backup and restore status reporting is more accurate, including a completedAt timestamp and correct failure handling for in-cluster restores.

Defaults and usability

  • Query statistics (pg_stat_statements) are now enabled by default, so slow-query insight needs no extra setup.
  • pg_cron and pg_repack extension toggles let you schedule jobs and reclaim bloat without manual SQL.
  • Client access rules can be declared with a pgHBA parameter instead of hand-editing cluster configuration.
  • Instances are validated at admission time, catching misconfigurations before they reach a running cluster.

These changes are available now. To learn more, visit the v0.3.4 release notes.