Is it possible without downtime? I have a 4 node galera multimaster cluster with haproxy, but the vanilla galera using 5.5.29 mysql and there is a bug which was busted in 5.5.31. Is it possible to stop node4, install percona xtradb, and join back, and the same on node3,2,1 ?
Thanks in advance for all hints.
I don’t see any reason why such approach should not work. However it’s always better to test the upgrade on some testing environment, like spin up just 2-3 virtual boxes nodes and try the same kind of upgrade.
Yep, I did that with success, they are synced, and work together. Thanks for the reply!