With cluster in version 5.6 with auto-crash recovery, I’m looking at one of the support scripts: Percona-XtraDB-Cluster-5.6.24-72.2/support-files/mysql.server.sh …
As it stands, if you have a cluster that died from a power cut, that script would prevent it from starting and being able to take advantage of auto-recovery. My question (as the topic states) - is there a reason to just bail if theres a pid file, but no matching process for it ? Why not just continue the start up ?
The function is “check_running()” and place in question is line 308 & 309.
If you go to the start case- line 339, abort unless ext_status=3 …
Any thoughts appreciated.