IN a 3 node percona galera cluster, I had started the first node with systemctl start mysql@bootstrap, and later two other nodes are started are started with systemctl start mysql.
Mysql service is running on all 3 nodes.
But when I execute systemctl mysql status on first node [where bootstrap] is executed, I get mysql service is not running.
systemctl mysql@bootstrap shows mysql as UP. Cant we get the status of mysql service from bootstrapped node?
Old thread i know but i think this is still current in 5.7.
How does xtradb behave on MySQL 8 now?
I have to say i think this bootstrapping change for systemd is a truly rubbish workaround.
We have this product in Live use since 2014 (5.5) and it has been very reliable and okay to support.
It is not ideal from CAP perspective but we have managed and it has served us really well in terms of performance and availability. We favour availability over consistency so that is where it falls down.
One stalling issue in 5.6.30 on FTS syncs.
We are finally getting CentOS7 into production and i am having to document how to support the “new” bootstrapping.
I cannot begin to describe how much trouble this is going to cause with admins running service stop and the process still running while they administer backups or other maintenance.
Not to mention it is confusing for me having worked with the cluster for years, like atluri007 says even a simple “status” misses a running process.
I see mariadb using systemctl set-environment. This seems much more workable but i cannot get it to work with bootstrap-pxc or wsrep-new-cluster.
I know this is a systemd problem but there must be a better workaround than this from a production support point of view.
Happy to talk more or shut up as i expect this will be the death of PXC for us.
Will continue to use the great Percona products but a simple managed master and slave with MHA will have to do.