I have a 3 node PXC 5.7 cluster which has been running for some time and recently needed to increase the innodb_log_file_size. I gracefully stopped node 3 and applied the change to innodb_log_file_size then started it back up. It successfully resized the log files when it restarted (thanks to the automatic resize feature in 5.6) and rejoined the cluster after performing an IST.
I do plan to make the change on the other 2 nodes at some point but are there any risks in running with mismatched values for innodb_log_file_size? For example if a node crashes and requires a SST from a donor node, will the mismatched innodb_log_file_size cause any problems? I’ve tried to search for any mention of this sort of configuration but didn’t find anything for or against it.