Why do helm charts disable persistence for elasticsearch?

Upon upgrading to elasticsearch 7.6.2, persistence was disabled. Why is that? The elasticsearch charts docs seem to suggest that this would not be a correct configuration for data nodes. And indeed when I tried to move a pod to a new compute node the cluster cannot come online (/_cluster/health shows an unassigned shard), presumably because its data is gone?

I have never used elasticsearch before so probably I am misunderstanding something.


For our helm chart, the temporal services themselves are set up in a production quality way. But we only recommend deploying the dependencies included in our helm chart for testing and development.

In this case if a user is deploying to a local kubernetes cluster using kind, more work would need to be done to enable persistent volumes. It’s another thing that can go wrong for people just trying to test something or develop against temporal.

That said, it’s certainly possible to set elasticsearch.persistence.enabled to true for your helm installation if you need it in your values file or on the command line for your helm install.