Rolling restart Kafka Connect workers

You can initiate a rolling restart of your Kafka Connect workers by annotating the StrimziPodSet resource or the individual Kafka Connect pods with the strimzi.io/manual-rolling-update="true" annotation. You annotate the StrimziPodSet if you want to restart all workers in your cluster. You annotate individual pods if you want to restart specific workers.

Annotating the StrimziPodSet resource

Each Kafka Connect cluster has their own StrimziPodSet resource. This resource manages all the pods related to the Kafka Connect cluster. Annotating this resource restarts all workers in your cluster.

kubectl annotate strimzipodset [***CONNECT CLUSTER NAME***]-connect strimzi.io/manual-rolling-update="true" --namespace [***NAMESPACE***]

In the next reconciliation loop, the Strimzi Cluster Operator initiates a rolling restart of all pods. After the pods are restarted, the annotation is automatically removed from the StrimziPodSet.

Annotating Pod resources

If you want to restart a specific pod or a set of specific pods, you annotate each Pod resource individually.

kubectl annotate pod [***KAFKA CONNECT POD***] strimzi.io/manual-rolling-update="true" --namespace [***NAMESPACE***]

The annotated pod is rolling restarted with the next reconciliation loop. The annotation is automatically removed from the pod after the pod is restarted.