Changelog 2025-11-04

PostgreSQL with CloudNativePG (CNPG) can be replicated. You can try it out by setting the number of instances to 2 or 3. For more details, refer to PostgreSQL replication.

By default, Envoy sidecar proxy is disabled on PostgreSQL instances. You can enable it by setting spec.parameters.service.enableEnvoy to true. When enabled, Envoy provides an additional proxy layer that can be used for advanced monitoring and observability purposes. More details in the Envoy sidecar proxy .

By default, StackGres automatically distributes the specified resources across all containers (including Patroni and sidecars). However, you can enable dedicated resource allocation for Patroni by setting spec.parameters.service.dedicatedPatroniResources to true. More details in the Dedicated patroni resources.

We have updated MariaDB collation defaults from utf8mb3_unicode_ci to utf8mb4_unicode_ci. The current defaults were outdated. All new MariaDB instances will be created with default collation utf8mb4_unicode_ci, see Create MariaDB. For troubleshooting collation settings refer to Collation issues.

We have identified and resolved a bug in AppCat where custom instance sizes were not correctly aligning requests with the configured limits. After the update, your instances will have slightly higher resource reservations, but this ensures guaranteed Quality of Service (QoS) for your applications. See Pod Quality of Service Classes for more information. If you prefer to maintain lower request values than your limits, you can easily adjust this by following our custom sizing documentation Plans and Sizing. This only affects instances which have custom CPU and memory limits without specific request specified.