hi, we are trying to upgrade our cluster from .29 to .36. The old entrypoint script in .29 only did the init checks if the datadir is empty whereas the new one also compares the versions. In our case this leads to triggering the 120s timeout here: percona-docker/percona-xtradb-cluster-8.0/dockerdir/entrypoint.sh at main · percona/percona-docker · GitHub
As our cluster is quite large with 1,3 TB data the startup takes some time and i wondered what the reasons behind this change were. I understand it if we start a new instance that we can predict the time it should take but with a version upgrade the time is probably very different depending on the scenario. This leads to a full stop of the mysql process while in the upgrade procedure of the underlying databases. Or am i missing a changed procedure for doing the upgrade with tthe container version? Couldn’t find any infos on this topic in the readme. Our update strategy was to do a maintenance perdiod, stop the whole cluster, update the bootstrap node and join the other nodes (with a new version using a full sst).
Thanks
Daniel