Recreate a cluster with the same cluster_uuid in k8s stetefullset configuration

Elastic Stack Elasticsearch
1

Hello , i'm able to upgrade eleasticsearch from v 6.6.0 to 7.1.1 using the officially provided docker in a stetefull set in kubernetes without a problem. After applying the cluster.initial_master_nodes with a list of 3 hosts the master is elected and everything works fine. After i redeploy the same version the pods are recreated and they complain about being part of the previous cluster ( referencing the old cluster_uuid ) and that they cannot form a new cluster .
The setup is: 3 pods / nodes acting as masters + 2 data nodes
My questions are the following :

  1. I would like to avoid storing the /data folder from the master nodes ( pods ) on a persistent volume, so is it possible to inject the same cluster_uuid into the configuration when recreating the masters , or maybe i'll need to pass more metadata ?

  2. Since version 7 i can see that there are a lot of changes in cluster formation. Can you share what are the best practices in using the kubernetes environment and keeping stuff agile and stable at the same time ?

Thank you !

2

It is essential that the master nodes have persistent storage. From the docs:

Master nodes must have access to the data/ directory (just like data nodes) as this is where the cluster state is persisted between node restarts.

2 Likes
3

Thanks David for your quick reply, i'll change the config and try to keep the data folder.

4

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