Index template lost after recreate es cluster in Kubernetes environment

Hi, All

I'm using Kubernetes to deploy es cluster and I'm really stuck at stability. (Referenced https://github.com/pires/kubernetes-elasticsearch-cluster)
I have 3 master node and 2 client node deployed as stateless pods (without data persists), and 3 data nodes deployed as statefulset (data persists).

When I first started the cluster, everything is fine. But after delete the whole cluster and start again, all index patterns are lost and all custom templates are gone, I have to re-add them back. So anyone can tell me anything else should I to in Kubernetes to avoid data lose? Thanks a lot.

Dedicated master nodes do not store data, but still need persistent storage for the cluster state, so I believe you should make them stateful as well.

I will try, and how nodes are managed by elasticsearch? I mean, re-creating pods is something like creating new machines, all of the nodes are not the old ones anymore, so uuid may be changed, will this impact something?

I combined data node and master node, now everything looks fine. Any performance concerns for doing this?

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.