Hello all,
We are using:
PXC galera stateful state
Ws-rep
PVC 100g x 3
with MySql on K8S
Using method Multi master for each pod
When I am killing a pod from the K8S UI, in order to test High availability, by the time that the POD restart to return to the cluster as the 3[SUP]rd[/SUP] pod, it (guessing) hold corrupted data, and for some unknown reason Sync with the remain healthy PODS, his corrupted data, leaving the entire Cluster down.
Does any one knows why it is happening and have a solution for it ?
Hello there could you upload some additional information for the team here, please? If your issue is very specific to your environment there may be a limit to what they can help with via the community forum, but let’s see how we go?
Please upload:
[LIST]
[]log files inc error logs for each of the nodes
[]my.cnf for each of the nodes
[]explicit versions of Percona XtraDB Cluster software and OS
[]a bit about the circumstances e.g. is this a new environment, one that’s been working well for a while, have you recently upgraded etc.
[/LIST] Basically anything around this that might help pinpoint the problem. Once you’ve replied I’ll let the team know you have updated.