I have a Redis cluster of 6 nodes, running in my Kubernetes cluster as a stateful set. Since it was for testing and not yet on production, all of the Redis nodes were on the same machine. Of course, the machine failed, and all of Redis' nodes crashed immediately. When the machine was back alive the pods were recreated and was given different cluster ips, therefore they couldn't re-connect with each other.
I need to find a solution for a disaster case such as this. Let's assume all the nodes were reassigned with different ips, how can I configure the nodes to get to other ips?
The slaves are easy to reset with the CLUSTER RESET command, but the masters contain slots and data that shouldn't be deleted.
Should I manually re-write nodes.conf? I'm afraid this will make it even worse? I there a known method to deal with it?
Thanks!