Plesk 12.5 with MariaDB 10

[Oliver Meyer]

Hi,

last weekend I upgraded one of my servers from Plesk 12 to 12.5 and encountered a problem afterwards.

OS: CentOS 7.1.1503 x86_64
Plesk: 12.5.30 MU3
MySQL: MariaDB 10.0.21 (official MariaDB repo for CentOS 7)

Problem:

MariaDB installed from the official MariaDB repo is not using a systemd-service but an old-style SysV init-script (/etc/init.d/mysql).
Now, after upgrading to Plesk 12.5 I'm not able to start / restart the plesk-services, because they expect an existing MariaDB systemd-service:

[root@s1 ~]# systemctl restart sw-cp-server.service
Failed to issue method call: Unit mariadb.service failed to load: No such file or directory.

[root@s1 ~]# systemctl restart sw-engine.service
Failed to issue method call: Unit mariadb.service failed to load: No such file or directory.

[root@s1 ~]# systemctl restart psa.service
Failed to issue method call: Unit mariadb.service failed to load: No such file or directory.

------

Can this be fixed with a forthcoming MU or is there a workaround?

Best regards,

Oliver Meyer

 

[mizar]

Hello,

It seems psa.service require mariadb.service, and it absense confuses systemd. As workaround you can edit your plesk systemd unit (/lib/systemd/system/psa.service) in this way: remove mariadb.service from Requires= and add Wants=mariadb.service mysql.service and also add mariadb.service mysql.service. Finally your unit should looks like such as:

[Unit]
Description=Plesk Panel
Requires=sw-cp-server.service sw-engine.service
Wants=mariadb.service mysql.service
After=network.target mariadb.service mysql.service

[Service]
Type=oneshot
RemainAfterExit=yes
EnvironmentFile=-/etc/sysconfig/psa
ExecStart=/usr/lib64/plesk-9.0/psa_service execute_actions

[Install]
WantedBy=multi-user.target

BTW I added the bug report for it, PPP-17732 just for reference

 

[Oliver Meyer]

If someone else want to use this workaround too, please don't forget to reload systemd after the changes.

systemctl daemon-reload

 

[Reqlez]

yea i just noticed this as well, but with me its the watchdog that informed me :

ct 1 23:09:27 http01 monit: 'mysql' failed to start

Oct 1 23:09:28 http01 monit: 'mysql' process is not running

Oct 1 23:09:28 http01 monit: 'mysql' trying to restart

Oct 1 23:09:28 http01 monit: 'mysql' start: /usr/sbin/service

Oct 1 23:09:28 http01 monit: Redirecting to /bin/systemctl start mariadb.service

Oct 1 23:09:28 http01 monit: Failed to issue method call: Unit mariadb.service failed to load: No such file or directory.

 

[Reqlez]

Oct 2 01:12:27 http01 monit: Starting MySQL SUCCESS!

After applying the patch form other post, at least watchdog seems like it can start the service, but it still reporting service restarting too many times so it drops from monitoring.

 

[mizar]

You should change start/stop command for MySQL server in watchdog:
1. Stop monitoring
2. Click in MySQL in watchdog services list
3. Replace start command to /usr/bin/systemctl start mysql, and stop to /usr/bin/systemctl stop mysql
4. Click OK, and start monitoring

 

[Reqlez]

You should change start/stop command for MySQL server in watchdog:
1. Stop monitoring
2. Click in MySQL in watchdog services list
3. Replace start command to /usr/bin/systemctl start mysql, and stop to /usr/bin/systemctl stop mysql
4. Click OK, and start monitoring

Yup, it worked ! Thanks ! Maybe its time to start running some websites and find more work for you guys ! work = bugs ;-)