Release notes for CloudNativePG 1.24

History of user-visible changes in the 1.24 minor release of CloudNativePG.

For a complete list of changes, please refer to the commits on the release branch in GitHub.

Version 1.24.4

Release date: May 23, 2025

Important Changes

• CloudNativePG is now officially a CNCF project: CloudNativePG has been accepted into the Cloud Native Computing Foundation (CNCF), marking a significant milestone in its evolution. As part of this transition, the project is now governed under CloudNativePG, a Series of LF Projects, LLC, ensuring long-term sustainability and community-driven innovation. (#7203)

Enhancements

• Added the KUBERNETES_CLUSTER_DOMAIN configuration option to the operator, allowing users to specify the domain suffix for fully qualified domain names (FQDNs) generated within the Kubernetes cluster. If not set, it defaults to cluster.local. (#6989)

• Implemented the cnpg.io/validation annotation, enabling users to disable the validation webhook on CloudNativePG-managed resources. Use with caution, as this allows unrestricted changes. (#7196)

• Added support for collecting pg_stat_wal metrics in PostgreSQL 18. (#7005)

• Added support for LZ4, XZ, and Zstandard compression methods when archiving WAL files via Barman Cloud (deprecated). (#7151)

Security

• Set imagePullPolicy to Always for the operator deployment to ensure that images are always pulled from the registry, reducing the risk of using outdated or potentially unsafe local images. (#7250)

Fixes

• Fixed native replication slot synchronization and logical replication failover for PostgreSQL 17 by appending the dbname parameter to primary_conninfo in replica configurations (#7298).

• Improved backup efficiency by introducing a fail-fast mechanism in WAL archiving, allowing quicker detection of unexpected primary demotion and avoiding unnecessary retries (#7483).

• Fixed an off-by-one error in parallel WAL archiving that could cause one extra worker process to be spawned beyond the requested number (#7389).

• Resolved a race condition that caused the operator to perform two switchovers when updating the PostgreSQL configuration. (#6991)

• Corrected the PodMonitor configuration by adjusting the matchLabels scope for the targeted pooler and cluster pods. Previously, the matchLabels were too broad, inadvertently inheriting labels from the cluster and leading to data collection from unintended targets. (#7063)

• Added a webhook warning for clusters with a missing unit (e.g., MB, GB) in the shared_buffers configuration. This will become an error in future releases. Users should update their configurations to include explicit units (e.g., 512MB instead of 512). (#7160)

• CloudNativePG Interface (CNPG-I):

• Implemented automatic reloading of TLS certificates for plugins when they change. (#7029)

• Ensured the operator properly closes the plugin connection when performing a backup using the plugin. (#7095, #7096)

• Improved performance and resilience of CNPG-I by removing timeouts for local plugin operations, avoiding failures during longer backup or WAL archiving executions (#7496).

• cnpg plugin:

• Ensured that the primary Pod is recreated during an imperative restart when primaryUpdateMethod is set to restart, aligning its definition with the replicas. (#7122)

Changes

• Updated the default PostgreSQL version to 17.5 for new cluster definitions. (#7556)

• Updated the default PgBouncer version to 1.24.1 for new Pooler deployments (#7399).

Version 1.24.3

Release Date: February 28, 2025

Enhancements

• Introduced a startup probe for the operator to enhance reliability and prevent premature liveness probe failures during initialization. (#7008)
• Added support for using the -r service with the Pooler. (#6868)
• Introduced an optional --ttl flag for the pgbench plugin, enabling automatic deletion of completed jobs after a user-defined duration. (#6701)
• Marked known error messages from the Azure CSI Driver for volume snapshots as retryable, improving resilience. (#6906)
• Updated the default PostgreSQL version to 17.4 for new cluster definitions. (#6960)

Security

• The operator image build process has been enhanced to strengthen security and transparency. Images are now signed with cosign, and OCI attestations are generated, incorporating the Software Bill of Materials (SBOM) and provenance data. Additionally, OCI annotations have been added to improve traceability and ensure the integrity of the images.

Bug Fixes

• Fixed inconsistent behavior in default probe knob values when .spec.probes is defined, ensuring users can override all settings, including failureThreshold. If unspecified in the startup probe, failureThreshold is now correctly derived from .spec.startupDelay / periodSeconds (default: 10, now overridable). The same logic applies to liveness probes via .spec.livenessProbeTimeout. (#6656)
• Managed service ports now take precedence over default operator-defined ports. (#6474)
• Fixed an issue where WAL metrics were unavailable after an instance restart until a configuration change was applied. (#6816)
• Fixed an issue in monolithic database import where role import was skipped if no roles were specified. (#6646)
• Added support for new metrics introduced in PgBouncer 1.24. (#6630)
• Improved handling of replication-sensitive parameter reductions by ensuring timely reconciliation after primary server restarts. (#6440)
• Introduced a new isWALArchiver flag in the CNPG-I plugin configuration, allowing users to designate a plugin as a WAL archiver. This enables seamless migration from in-tree Barman Cloud support to the plugin while maintaining WAL archive consistency. (#6593)
• Ensured override.conf is consistently included in postgresql.conf during replica cluster bootstrapping, preventing replication failures due to missing configuration settings. (#6808)
• Ensured override.conf is correctly initialized before invoking pg_rewind to prevent failures during primary role changes. (#6670)
• Enhanced webhook responses to return both warnings and errors when applicable, improving diagnostic accuracy. (#6579)
• Ensured the operator version is correctly reconciled. (#6496)
• Improved PostgreSQL version detection by using a more precise check of the data directory. (#6659)
• Volume Snapshot Backups:
  • Fixed an issue where unused backup connections were not properly cleaned up. (#6882)
  • Ensured the instance manager closes stale PostgreSQL connections left by failed volume snapshot backups. (#6879)
  • Prevented the operator from starting a new volume snapshot backup while another is already in progress. (#6890)
• cnpg plugin:
  • Restored functionality of the promote plugin command. (#6476)
  • Enhanced kubectl cnpg report --logs <cluster> to collect logs from all containers, including sidecars. (#6636)
  • Ensured pgbench jobs can run when a Cluster uses an ImageCatalog. (#6868)

Technical Enhancements

• Added support for Kubernetes client-gen, enabling automated generation of Go clients for all CloudNativePG CRDs. (#6695)

Version 1.24.2

Release Date: December 23, 2024

Enhancements

• Enable customization of startup, liveness, and readiness probes through the .spec.probes stanza. (#6266)
• Add the cnpg.io/userType label to secrets generated for predefined users, specifically superuser and app. (#4392)
• Improved validation for the spec.schedule field in ScheduledBackups, raising warnings for potential misconfigurations. (#5396)
• cnpg plugin:
  • Honor the User-Agent header in HTTP requests with the API server. (#6153)

Bug Fixes

• Ensure the former primary flushes its WAL file queue to the archive before re-synchronizing as a replica, reducing recovery times and enhancing data consistency during failovers. (#6141)
• Clean the WAL volume along with the PGDATA volume during bootstrap. (#6265)
• Update the operator to set the cluster phase to Unrecoverable when all previously generated PersistentVolumeClaims are missing. (#6170)
• Fix the parsing of the synchronous_standby_names GUC when .spec.postgresql.synchronous.method is set to first. (#5955)
• Resolved a potential race condition when patching certain conditions in CRD statuses, improving reliability in concurrent updates. (#6328)
• Correct role changes to apply at the transaction level instead of the database context. (#6064)
• Remove the primary_slot_name definition from the override.conf file on the primary to ensure it is always empty. (#6219)
• Configure libpq environment variables, including PGHOST, in PgBouncer pods to enable seamless access to the pgbouncer virtual database using psql from within the container. (#6247)
• Remove unnecessary updates to the Cluster status when verifying changes in the image catalog. (#6277)
• Prevent panic during recovery from an external server without proper backup configuration. (#6300)
• Resolved a key collision issue in structured logs, where the name field was inconsistently used to log two distinct values. (#6324)
• Ensure proper quoting of the inRoles field in SQL statements to prevent syntax errors in generated SQL during role management. (#6346)
• cnpg plugin:
  • Ensure the kubectl context is properly passed in the psql command. (#6257)
  • Avoid displaying physical backups block when empty with status command. (#5998)

Version 1.24.1

Release date: Oct 16, 2024

Enhancements:

• Remove the use of pg_database_size from the status probe, as it caused high resource utilization by scanning the entire PGDATA directory to compute database sizes. The kubectl status plugin will now rely on du to provide detailed size information retrieval (#5689).
• Add the ability to configure the full_page_writes parameter in PostgreSQL. This setting defaults to on, in line with PostgreSQL's recommendations (#5516).
• Plugin:
  • Add the logs pretty command in the cnpg plugin to read a log stream from standard input and output a human-readable format, with options to filter log entries (#5770)
  • Enhance the status command by allowing multiple -v options to increase verbosity for more detailed output (#5765).
  • Add support for specifying a custom Docker image using the --image flag in the pgadmin4 plugin command, giving users control over the Docker image used for pgAdmin4 deployments (#5515).

Fixes:

• Resolve an issue with concurrent status updates when demoting a primary to a designated primary, ensuring smoother transitions during cluster role changes (#5755).
• Ensure that replica PodDisruptionBudgets (PDB) are removed when scaling down to two instances, enabling easier maintenance on the node hosting the replica (#5487).
• Prioritize full rollout over inplace restarts (#5407).
• When using .spec.postgresql.synchronous, ensure that the synchronous_standby_names parameter is correctly set, even when no replicas are reachable (#5831).
• Fix an issue that could lead to double failover in cases of lost connectivity (#5788).
• Correctly set the TMPDIR and PSQL_HISTORY environment variables for pods and jobs, improving temporary file and history management (#5503).
• Plugin:
  • Resolve a race condition in the logs cluster command (#5775).
  • Display the potential sync status in the status plugin (#5533).
  • Fix the issue where pods deployed by the pgadmin4 command didn’t have a writable home directory (#5800).

Supported versions

• PostgreSQL 17 (PostgreSQL 17.0 is the default image)

Version 1.24.0

Release date: Aug 22, 2024

Important changes:

• Deprecate the role label in the selectors of Service and PodDisruptionBudget resources in favor of cnpg.io/instanceRole (#4897).
• Fix the default PodAntiAffinity configuration for PostgreSQL Pods, allowing a PostgreSQL and a Pooler Instance to coexist on the same node when the anti-affinity configuration is set to required (#5156).

Features:

• Distributed PostgreSQL Topologies: Enhance the replica cluster feature to create distributed database topologies for PostgreSQL that span multiple Kubernetes clusters, enabling hybrid and multi-cloud deployments. This feature supports:
  • Declarative Primary Control: Easily specify which PostgreSQL cluster acts as the primary in a distributed setup (#4388).
  • Seamless Switchover: Effortlessly demote the current primary and promote a selected replica cluster, typically in a different region, without needing to rebuild the former primary. This ensures high availability and resilience in diverse environments (#4411).
• Managed Services: Introduce managed services via the managed.services stanza (#4769 and #4952), allowing you to:
  • Disable the read-only and read services via configuration.
  • Leverage the service template capability to create custom service resources, including load balancers, to access PostgreSQL outside Kubernetes (particularly useful for DBaaS purposes).
• Enhanced API for Synchronous Replication: Introducing an improved API for explicit configuration of synchronous replication, supporting both quorum-based and priority list strategies. This update allows full customization of the synchronous_standby_names option, providing greater control and flexibility (#5148).
• WAL Disk Space Exhaustion: Safely stop the cluster when PostgreSQL runs out of disk space to store WAL files, making recovery easier by increasing the size of the related volume (#4404).

Enhancements:

• Add support for delayed replicas by introducing the .spec.replica.minApplyDelay option, leveraging PostgreSQL's recovery_min_apply_delay capability (#5181).
• Introduce postInitSQLRefs and postInitTemplateSQLRefs to allow users to define postInit and postInitTemplate instructions as one or more config maps or secrets (#5074).
• Add transparent support for PostgreSQL 17's allow_alter_system parameter, enabling or disabling the ALTER SYSTEM command through the .spec.postgresql.enableAlterSystem option (#4921).
• Allow overriding the query metric name and the names of the columns using a name key/value pair, which can replace the name automatically inherited from the parent key (#4779).
• Enhanced control over exported metrics by making them subject to the value returned by a custom query, which is run within the same transaction and defined in the predicate_query field (#4503).
• Allow additional arguments to be passed to barman-cloud-wal-archive and barman-cloud-wal-restore (#5099).
• Introduce the reconcilePodSpec annotation on the Cluster and Pooler resources to control the restart of pods following a change in the Pod specification (#5069).
• The readiness probe now fails for streaming replicas that were never connected to the primary instance, allowing incoherent replicas to be discovered promptly (#5206).
• Support the new metrics introduced in PgBouncer 1.23 in the Pooler metrics collector (#5044).
• cnpg plugin updates:
  • Enhance the install generate command by adding a --control-plane option, allowing deployment of the operator on control-plane nodes by setting node affinity and tolerations (#5271).
  • Enhance the destroy command to delete also any job related to the target instance (#5298).
  • Enhanced the status command to display demotionToken and promotionToken when available, providing more detailed operational insights with distributed topologies (#5149).
  • Added support for customizing the remote database name in the publication and subscription subcommands. This enhancement offers greater flexibility for synchronizing data from an external cluster with multiple databases (#5113).

Security:

• Add TLS communication between the operator and instance manager (#4442).
• Add optional TLS communication for the instance metrics exporter (#4927).

Fixes:

• Enhance the mechanism for detecting Pods that have been terminated but not deleted during an eviction process, and extend the cleanup process during maintenance windows to include unschedulable Pods when the reusePVC flag is set to false (#2056).
• Disable pg_rewind execution for newly created replicas that employ VolumeSnapshot during bootstrapping to avoid introducing a new shutdown checkpoint entry in the WAL files. This ensures that replicas can reconnect to the primary without issues, which would otherwise be hindered by the additional checkpoint entry (#5081).
• Gracefully handle failures during the initialization of a new instance. Any remaining data from the failed initialization is now either removed or, if it's a valid PostgreSQL data directory, moved to a backup location to avoid possible data loss (#5112).
• Enhance the robustness of the immediate backups reconciler by implementing retry logic upon initial backup failure (#4982).
• Wait for the postmaster to shut down before starting it again (#4938).
• Ensure that the Pooler service template can override the default service (#4846).
• Exclude immutable databases from pg_database metric monitoring and alerting processes (#4980).
• Removed unnecessary permissions from the operator service account (#4911).
• Fix cluster role permissions for ClusterImageCatalogs (#5034).
• Ensure the operator initiates a rollout of the Pooler instance when the operator image is upgraded (#5006)
• Address race condition causing the readiness probe to incorrectly show "not ready" after a PostgreSQL restart, even when the postmaster was accessible (#4920).
• Prevent reconciliation of resources that aren't owned by a Pooler (#4967).
• Renew the certificates managed by the operator when the DNS Subject Alternative Names (SANs) are updated (#3269, #3319).
• Set PVC default AccessModes in the template only when unspecified (#4845).
• Gracefully handle unsatisfiable backup schedule (#5109).
• Synchronous replication self-healing checks now exclude terminated pods, focusing only on active and functional pods (#5210).
• The instance manager will now terminate all existing operator-related replication connections following a role change in a replica cluster (#5209).
• Allow setting smartShutdownTimeout to zero, enabling immediate fast shutdown and bypassing the smart shutdown process when required (#5347).
• cnpg plugin:
  • Properly handle errors during the status command execution.
  • Support TLS in the status command (#4915).

Supported versions

• Kubernetes 1.31, 1.30, 1.29, and 1.28
• PostgreSQL 16, 15, 14, 13, and 12
  • PostgreSQL 16.4 is the default image
  • PostgreSQL 12 support ends on November 12, 2024