> For the complete documentation index, see [llms.txt](https://mariadb.com/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://mariadb.com/docs/release-notes/enterprise-operator/26.06/26.06.2.md).

# MariaDB Enterprise Kubernetes Operator 26.06.2

MariaDB Enterprise Operator 26.06.2 is a maintenance release of MariaDB Enterprise Operator, released on 2026-08-28

**Release date**: 28 August 2026

MariaDB Enterprise Kubernetes Operator 26.06.2 is a maintenance release for the 26.06 series. It focuses on the reliability of replication topologies during switchovers and updates.

If you are updating from a previous version, **please follow the** [**UPDATE GUIDE**](https://mariadb.com/docs/tools/mariadb-enterprise-operator/updates/update-26.06.2) to ensure a safe transition.

### Changelog

* **K8SDKR-304**: Multiple fixes in the failover and switchover mechanisms fully driven by the operator (not MaxScale):
  * `RESET MASTER;` is no longer issued when configuring a replica, in **any** replication topology. This is safer: binary logs and GTID positions are preserved when a primary is demoted, which keeps point-in-time recovery archival consistent and surfaces errant transactions instead of silently erasing them.
  * `Error 1948`, returned when resetting `gtid_slave_pos` while promoting a new primary, is now handled. Previously the error aborted the promotion and left the primary half configured.
  * `User`, `Grant` and `Database` resources are no longer reconciled while the `MariaDB` is in a transient state, such as an ongoing switchover.
* **K8SDKR-307**: `read_only` is now enforced on the replicas of clusters that do not use MaxScale even when the `MariaDB` resource is not `Ready`. Since `read_only` is runtime state only and is not persisted in the MariaDB configuration, a replica that restarted while unhealthy could stay writable indefinitely. Clusters with `spec.maxScaleRef` set are unaffected and behave exactly as before.
* **K8SDKR-297**: A rolling update no longer gets stuck on a MaxScale switchover when the promoted primary has an empty `gtid_binlog_pos`. The operator now forces the primary configuration in this scenario, which populates the binary log position and makes the node a valid demotion target for `mariadbmon`.
* **K8SDKR-303**: The `copy-agent` init container image now honors `updateStrategy.autoUpdateDataPlane`, like the rest of the [data-plane](https://mariadb.com/docs/tools/mariadb-enterprise-operator/topologies/data-plane) containers. Previously it was taken directly from the running operator image, so updating the operator triggered an unsolicited rolling restart of every `MariaDB` with replication enabled, even when `autoUpdateDataPlane` was `false`.
* **K8SDKR-296**: The downtime window observed during a failover with MaxScale has been decreased. See [MXS-6859](https://jira.mariadb.org/browse/MXS-6859).
* **K8SDKR-310**: New `replication.semiSyncWaitNoSlave` field in the `MariaDB` resource, which maps to [`rpl_semi_sync_master_wait_no_slave`](https://mariadb.com/docs/server/ha-and-performance/standard-replication/semisynchronous-replication#rpl_semi_sync_master_wait_no_slave). Using this field requires the [data-plane](https://mariadb.com/docs/tools/mariadb-enterprise-operator/topologies/data-plane) to be updated to `26.6.2`.
* **K8SDKR-177**: OpenShift `4.20` and `4.22` are now supported. The operator is certified and published in the `4.22`, `4.20`, `4.18` and `4.16` certified operator catalogs.

### Platform and component versions

The current release has been tested with the following versions:

| Platform/Component        | Version  |
| ------------------------- | -------- |
| Kubernetes                | 1.36     |
| OpenShift                 | 4.20     |
| MariaDB Enterprise Server | 11.8.8-5 |
| MaxScale                  | 25.10.3  |

<sub>*This page is: Copyright © 2026 MariaDB. All rights reserved.*</sub>

{% @marketo/form formid="4316" formId="4316" %}