Seeing that the nodes crash at the same seqno suggests (but not limited to) simple db inconsistency. In this case it is important to know how those nodes were initialized, is it a LAN or WAN cluster and was there any tinkering with pc.ignore_sb Galera parameter?
Seeing that the nodes crash at the same seqno suggests (but not limited to) simple db inconsistency. In this case it is important to know how those nodes were initialized, is it a LAN or WAN cluster and was there any tinkering with pc.ignore_sb Galera parameter?