Upgrade RisingWave in a Kubernetes cluster
This topic describes upgrade RisingWave in a K8s deployment with the RisingWave Kubernetes Operator and Helm.
CAUTION
When upgrading RisingWave, it’s important to be aware that there may be breaking changes. If you require technical support during the process of upgrading RisingWave in your production environments, please don’t hesitate to reach out to us.
NOTE
Assuming that the Kubernetes namespace is default
, if your RisingWave cluster is deployed in another namespace, please add the -n <namespace>
argument to the kubectl
and helm
commands below. Remember to replace the <namespace>
with your own namespace.
Upgrade RisingWave with Helm
- Check the current status of the RisingWave helm release.
The output should look like this:
- Update the local cache of available charts (packages) from the configured Helm chart repositories:
- Check for available versions of RisingWave Helm charts:
- Upgrade RisingWave. You can upgrade to the latest version or a particular version.
To upgrade to the latest version:
To upgrade to a particular version:
Upgrade RisingWave with the Operator
Remember to replace all the <risingwave-cluster>
with the real object name.
- Check the current status of the RisingWave cluster.
Expected output:
- Upgrade the image version. Remember to replace
<target-version>
with the target image version.
- Wait until all pods are recreated and the
Upgrading
condition becomesFalse
or empty.
If wait timeouts, please check if the pods are running properly.
- If any of the pods is pending or not running, you might need to troubleshoot first and see if there are problems with the pod itself. Possible issues include the image was not found or the pod operation was not scheduled.
- If the meta pod isn’t running and ready, please submit an issue with the logs and rollback by following the guides below.
- If the meta pod is running but other pods are not, please wait a minute and see if they will be running afterwards. Please submit an issue with the logs and roll back by following the instructions below.
Expected output:
Roll back when necessary
If any issue happens and it cannot be resolved within a certain timeframe, you can roll back the upgrade with the following command. Remember to replace the <version-before>
with the image version used before the upgrade. The other steps to verify the rollback are the same as the upgrade.
Was this page helpful?