Pause a MySQL Pod¶

Use the sleep-forever file in /var/lib/mysql when you need the MySQL container to stay up without mysqld, or when an unready Pod must report Ready so a rolling update can continue. See sleep-forever for what the file changes.

Use another container if mysql keeps crashing

The commands on this page use the mysql container (-c mysql). If this container crashes or restarts in a loop, kubectl exec into it can fail. The xtrabackup and logs containers of the same Pod mount the same data volume at /var/lib/mysql, so you can run the commands there instead. The xtrabackup container exists when backups are enabled (backup.enabled), and the logs container exists when persistent logging is enabled (logcollector.enabled). For example:

kubectl exec <cluster-name>-mysql-<N> -n <namespace> -c xtrabackup -- touch /var/lib/mysql/sleep-forever

Pause a Pod for debugging¶

Use the sleep-forever file to keep a MySQL Pod running without mysqld. This helps when mysqld crashes at startup and the container restarts in a loop, or when you need to inspect, repair, or copy files from the data directory while mysqld is stopped.

Create the file in the mysql container of the Pod:
```
kubectl exec <cluster-name>-mysql-<N> -n <namespace> -c mysql -- touch /var/lib/mysql/sleep-forever
```
If the mysql container restarts too often to run the command, run it in the xtrabackup or logs container instead. Those containers mount the same data volume. The file is on the Persistent Volume, so it stays in place after the next restart.
Delete the Pod. Kubernetes recreates it, and the container stops before starting mysqld:
```
kubectl delete pod <cluster-name>-mysql-<N> -n <namespace>
```

Check the container log for the node is going to infinity loop message:

kubectl logs <cluster-name>-mysql-<N> -n <namespace> -c mysql

Open a shell in the container and do your work. The data directory is available in /var/lib/mysql:
```
kubectl exec -it <cluster-name>-mysql-<N> -n <namespace> -c mysql -- bash
```
Remove the file to resume the Pod:
```
kubectl exec <cluster-name>-mysql-<N> -n <namespace> -c mysql -- rm /var/lib/mysql/sleep-forever
```
In a few seconds, the container exits and Kubernetes restarts it. The container then starts mysqld, and the Pod rejoins the cluster.

Unblock a stuck rolling update¶

When you change the Custom Resource or upgrade the database, the MySQL Pods restart one by one with the new configuration. Both the SmartUpdate and the RollingUpdate update strategies restart the next Pod only when all MySQL Pods are Ready. If one Pod is stuck in the unready state, for example because mysqld can’t join the cluster, the update stops and waits for this Pod. The update may be the fix for the Pod, but it never reaches the Pod.

The sleep-forever file makes the readiness probe pass right away, so the update can continue:

Find the Pod that isn’t Ready:

kubectl get pods -n <namespace> -l app.kubernetes.io/component=mysql

Create the file in the mysql container of this Pod:
```
kubectl exec <cluster-name>-mysql-<N> -n <namespace> -c mysql -- touch /var/lib/mysql/sleep-forever
```
The Pod becomes Ready with the next readiness probe, without a restart. The update continues. When it reaches this Pod, the Pod is recreated with the new configuration and doesn’t start mysqld.

Check that the Pod runs the new configuration. The two commands print the same value when it does:

kubectl get pod <cluster-name>-mysql-<N> -n <namespace> -o jsonpath='{.metadata.labels.controller-revision-hash}'
kubectl get sts <cluster-name>-mysql -n <namespace> -o jsonpath='{.status.updateRevision}'

Remove the file right away:
```
kubectl exec <cluster-name>-mysql-<N> -n <namespace> -c mysql -- rm /var/lib/mysql/sleep-forever
```
The container restarts, starts mysqld with the new configuration, and the Pod rejoins the cluster.

Warning

Remove the file as soon as the Pod runs the new configuration. With the SmartUpdate strategy, the Operator updates the primary Pod last. Before that, it switches the primary role to the first Ready secondary Pod. The Operator may choose the Pod with the sleep-forever file, and the switchover fails because mysqld doesn’t run there.

Last update: October 27, 2026
Created: October 27, 2026