Percona-cluster should show in status when mysql is down

[David Ames]

There are times when mysql (percona-cluster) has crashed but the juju status still shows Unit Ready.

Neither the percona-cluster charm nor the hacluster charm seem to recognize this.

Percona seems completely unaware if the service is actually running. asses_status should validate the functionality of the service.

The hacluster charm has the mysql monitor feature but it is not presenting that back to the end user via juju status or being in a blocked state.

[David Ames (thedac)]

Hacluster msyql monitor:

mysql_monitor(res_mysql_monitor)[178679]: ERROR: Not enough arguments [1] to ocf_log.
mysql_monitor(res_mysql_monitor)[178813]: MYSQL IS NOT RUNNING:
mysql_monitor(res_mysql_monitor)[178923]: MYSQL IS NOT RUNNING:

[Liam Young (gnuoy)]

The percona-cluster charm does seem to validate that mysql is running:

Unit Workload Agent Machine Public address Ports Message
percona-cluster/0* active idle 0 10.5.0.31 3306/tcp Unit is ready
  hacluster/0* active idle 10.5.0.31 Unit is ready and clustered
percona-cluster/1 blocked idle 1 10.5.0.3 3306/tcp Services not running that should be: mysql
  hacluster/1 active idle 10.5.0.3 Unit is ready and clustered
percona-cluster/2 active idle 2 10.5.0.72 3306/tcp Unit is ready
  hacluster/2 active idle 10.5.0.72 Unit is ready and clustered

I don't see an obvious commit since this bug was raised that would account for status being fixed but I don't seem to be seeing the issue. Am I missing something here?

[David Ames (thedac)]

This hacluster change added more robust status:
https://github.com/openstack/charm-hacluster/commit/93644400756170bb8e37d5a5f995a54a4a3bd8dd

This percona-cluster charm fix will also close this bug:
https://review.opendev.org/#/c/667753/4

[Alex Kavanagh (ajkavanagh)]

According to comment #3 this is now fixed as the associated patchset has been merged.