[BlueOnyx:20354] 5209R: MariaDB issues - please read

[Michael Stauber]

Hi all,

As you have seen on the list: The latest round of CentOS 7 updates
messed with MariaDB and caused a conflict with mariadb-galera.

As Galera was apparently not needed, I published a base-mysql-* update
that uninstalled it.

And that made matters worse.

Here is the problem:

We do have boxes with the "stock" MariaDB-5.5 from CentOS and we do have
5209R with the PKG MariaDB-10.1.18-1.

The difference:

MariaDB-10.1.18 has the start/stop script in the RPM mariadb-server.

The "stock" MariaDB-5.5 has the start/stop scripts in the RPM
mariadb-galera-server.

So uninstalling the Galera related RPMs breaks MariaDB-5.5 and it cannot
be restarted.

Long story short: Our update made things worse. Here is the fix:

yum clean all
yum update -y
swatch
/usr/sausalito/swatch/bin/check-mysql.sh

Then check if MariaDB is running:

systemctl status mariadb

It should be running again at that point. The run of
"/usr/sausalito/swatch/bin/check-mysql.sh" should only necessary if
MySQL monitoring is disabled in Active Monitor and this script does the
actual work of bringing the galera and mariadb-galera-* RPM's back.

In an unattended automated update the simply "yum update" run by the
cronjob should fix this entirely once Swatch runs *and* MySQL monitoring
is enabled in Active Monitor. By default it is.

My apologies for the mess. :-/

-- 
With best regards

Michael Stauber