KubeMQ
Release notes

Release 3.3.0

KubeMQ 3.3.0 starts only on supported installations, lets you choose the first dashboard administrator on Docker, and needs at least 3 servers on Kubernetes.

Released September 23, 2026. This release refuses installation shapes KubeMQ does not support and makes the first minutes of an evaluation easier to follow. It is the first release numbered above the legacy product; see How KubeMQ versions work.

Artifacts

ArtifactVersion
Server imageeurope-docker.pkg.dev/kubemq/images/kubemq-next:v3.3.0
Server image (FIPS)europe-docker.pkg.dev/kubemq/images/kubemq-next-fips:v3.3.0
Operator imageeurope-docker.pkg.dev/kubemq/images/kubemq-operator-next:v3.3.0, then v3.3.1 the same day
Helm chart kubemq-next3.3.0 (operator v3.3.0, server v3.3.0), then 3.3.1 (operator v3.3.1, server v3.3.0)
kmqNo new release

Before you upgrade

  • Only supported installations start. The server refuses to start as a native process, under a container runtime other than Docker, as a clustered Docker server, or on Kubernetes without the operator or with fewer than 3 servers. It stops before it reads the license or changes stored data. Move such an installation to a supported setup before you upgrade. Single-server Podman is supported from 3.5.0.
  • Kubernetes clusters need at least 3 servers. The operator refuses to reconcile a cluster with standalone: true or fewer than 3 servers, and leaves its running pods and volumes as they are. Move such a cluster before you upgrade. A new cluster also needs a license that covers its server count.
  • Kafka exposed outside a cluster needs per-server addresses. On a clustered KubemqCluster, spec.kafka.expose set to NodePort or LoadBalancer now requires spec.kafka.peers. The operator refuses the change before it touches the running cluster.

Changes

  • First administrator on Docker. On a fresh Docker server, management sign-in is on unless you turn it off, and the first person to open the dashboard chooses the administrator. Until then the dashboard and the management API stay locked; messaging keeps its own security settings. The choice survives restarts and container replacement, and recovering the administrator does not delete messages. Sign-in settings you set yourself are kept.
  • The first minutes say what is happening. A server started without a license key that cannot reach the license service now logs what it is waiting for and how to proceed. A Kafka connector you enabled by name fails startup when its port is taken, naming the port; the default-on Kafka connector is switched off instead, and the log and the dashboard banner say why.
  • Two licensing terms. Messages and the dashboard now say "evaluation" for the 14-day run without a key and "trial key" for the 30-day key from the trial form. When an evaluation ends, the message offers both ways forward: add a trial key, or remove the volume, which deletes its messages, and start again.
  • Dashboard. Empty lists are told apart from failed requests, with a retry. All six dashboard languages are complete.
  • RabbitMQ (AMQP 0-9-1). Clearing expired messages at the head of a queue no longer loses the live message behind them, and binding destinations are validated.
  • Kubernetes. On OpenShift, chart 3.3.0 no longer grants the server's service account the privileged security context constraint. Operator 3.3.1 changes only the operator's own install manifest, which no longer creates a Role for that constraint; clusters behave as with 3.3.0.

Known issues

  • Images are published for linux/amd64 only. Release 3.5.2 adds linux/arm64.
  • No default connector image is published. Set spec.image on every KubemqConnector.

Was this page helpful?

On this page