Issue Updating MariaDB from 10.5 to 10.11.4 breaks everything.

[MHC_1]

Server operating system version

AlmaLinux 9.6 (Sage Margay)

Plesk version and microupdate number

Obisidan 18.0.73

Plesk Obsidian 18.0.73 on AlmaLinux 9.6 (Sage Margay)

So we used the Plesk Update system and took our MariaDB 10.5 to 10.11.14 and now everything is broken.

The Plesk system auto-logged out during the update process; while hanging on doing something with psa.longtasks . However, the Repair utility did seem to suggest everything was working ok.

Plesk GUI seemed to be working ok.
Plesk PHPMyAdmin was loading DBs and tables ok .

But:
Websites can no longer connect to the database on localhost.
IDEs can no longer Tunnel into the database as before; with the issue

"DBMS: MariaDB (no ver.)Case sensitivity: plain=mixed, delimited=exactSocket fail to connect to host:address=(host=localhost)(port=3306) (type=primary). Connection refused."

And client webmail does not work;

"Oops... something went wrong!
An internal error has occurred. Your request cannot be processed at this time.
For administrators: Please check the application and/or server error logs for more information."

We have looked for log files but can't see any useful logs. Where are they kept?

 

[MHC_1]

UPDATE: restarting the server results in failure of Plesk to load:

[2025-10-28 13:58:34+00:00] INFO: Failed to start the database service. See /var/log/plesk/rc_actions.log and the service log for details.
[2025-10-28 13:58:34+00:00] ERROR: Could not start the database service.

 

[alapshin]

Hello @MHC_1
I suggest you the following solution Database connections initiated using PHP fail when SELinux is enabled on a Plesk for Linux server
In short

1. Restore SELinux context for MariaDB file:
   restorecon /usr/sbin/mariadbd
2. Restart MariaDB service:
   systemctl restart mariadb

If it will not help submit a request to Plesk support they should help you in place.

 

[MHC_1]

Hello @MHC_1
I suggest you the following solution Database connections initiated using PHP fail when SELinux is enabled on a Plesk for Linux server
In short

1. Restore SELinux context for MariaDB file:
   restorecon /usr/sbin/mariadbd
2. Restart MariaDB service:
   systemctl restart mariadb

If it will not help submit a request to Plesk support they should help you in place.

Thank you. However,
Rc_actions.log states:

Various Plugins seem to not be running or disabled. ffs.

INFO: [Tue Oct 28 14:07:37 UTC 2025]: Service: mysql, Action: start
Trying to start service mariadb... inactive
Oct 28 14:07:37 mhccorefour.co.uk mariadbd[6142]: 2025-10-28 14:07:37 0 [Note] InnoDB: Retry with innodb_force_recovery=5
Oct 28 14:07:37 mhccorefour.co.uk mariadbd[6142]: 2025-10-28 14:07:37 0 [ERROR] InnoDB: Plugin initialization aborted with error Data structure corruption
Oct 28 14:07:37 mhccorefour.co.uk mariadbd[6142]: 2025-10-28 14:07:37 0 [Note] InnoDB: Starting shutdown...
Oct 28 14:07:37 mhccorefour.co.uk mariadbd[6142]: 2025-10-28 14:07:37 0 [ERROR] Plugin 'InnoDB' registration as a STORAGE ENGINE failed.
Oct 28 14:07:37 mhccorefour.co.uk mariadbd[6142]: 2025-10-28 14:07:37 0 [Note] Plugin 'FEEDBACK' is disabled.
Oct 28 14:07:37 mhccorefour.co.uk mariadbd[6142]: 2025-10-28 14:07:37 0 [ERROR] Unknown/unsupported storage engine: InnoDB
Oct 28 14:07:37 mhccorefour.co.uk mariadbd[6142]: 2025-10-28 14:07:37 0 [ERROR] Aborting
Oct 28 14:07:37 mhccorefour.co.uk systemd[1]: mariadb.service: Main process exited, code=exited, status=1/FAILURE
Oct 28 14:07:37 mhccorefour.co.uk systemd[1]: mariadb.service: Failed with result 'exit-code'.
Oct 28 14:07:37 mhccorefour.co.uk systemd[1]: Failed to start MariaDB 10.11.14 database server.

 

[alapshin]

Thank you for providing log snippets. The issue is in InnoDB files corruption (most probably).

Try the following soluiton MySQL/MariaDB fails to start: Upgrade after a crash is not supported

1. Rename the corrupted ib_logfile files:
   mv /var/lib/mysql/ib_logfile0 /var/lib/mysql/ib_logfile0.bak
   mv /var/lib/mysql/ib_logfile1 /var/lib/mysql/ib_logfile1.bak
2. Start MySQL/MariaDB server:
   systemctl start mariadb

Also try to repair files using repair kit:
Log in to its interface by https://<IP>:8443/repair and using adminnistrator credentials
Then check file permissions and InnoDB files:


Contact submit a Plesk support for deeper investigation if the solution will not helped.

 

[Tofandel]

Note that if you do this, you will be able to recover and start mariadb, but you won't have any data. You should restore it from the backup before the upgrade that you will find in /var/lib/psa/dumps/mysql.preupgrade.*.dump.gz (look up based on your version)

There is ways to recover from this error by starting mariadb in safe mode and fixing the corruption, because it likely only is in a few tables, but it's an advanced process. I did have to do it once and successfully recovered

I would suggest before going the backup route to add to your mysql conf
[mysqld]
innodb_force_recovery = 1

Try to restart mariadb and if it fails increase from 2 to 5 until it does start

 

[Tofandel]

If you don't have a full backup and get it to start with the above config, dump everything with `mysqldump --all-databases > backup.sql`, remove the innodb_force_recovery config, stop mariadb, run
```
mv /var/lib/mysql/ibdata1 /var/lib/mysql/ibdata1.bak
mv /var/lib/mysql/ib_logfile0 /var/lib/mysql/ib_logfile0.bak
mv /var/lib/mysql/ib_logfile1 /var/lib/mysql/ib_logfile1.bak
```

Restart mariadb and reimport your dump `mysql < backup.sql`

There is a way to check in the logs, which tables are corrupted and only dump those and reimport only those without rebuilding the whole database if you have a really big database but I won't get into it here

 

[MHC_1]

Ok.
Thank you both for your help.

Yes, There seems to be corruption on ib_logfile0 . Does this file get saved on any system wide backups? I have a remote backup that's ~92Gb and wondering if I can simply download this and reinstate the previously working ib_logfile0 into /var/lib/mysql/ ?

 

[MHC_1]

Hello @MHC_1
I suggest you the following solution Database connections initiated using PHP fail when SELinux is enabled on a Plesk for Linux server
In short

1. Restore SELinux context for MariaDB file:
   restorecon /usr/sbin/mariadbd
2. Restart MariaDB service:
   systemctl restart mariadb

If it will not help submit a request to Plesk support they should help you in place.

Thank you. I have completed this fix. So once the ib_logfile0's are resolved hopefully the system will load correctly !

 

[MHC_1]

Thank you for providing log snippets. The issue is in InnoDB files corruption (most probably).

Try the following soluiton MySQL/MariaDB fails to start: Upgrade after a crash is not supported

1. Rename the corrupted ib_logfile files:
   mv /var/lib/mysql/ib_logfile0 /var/lib/mysql/ib_logfile0.bak
   mv /var/lib/mysql/ib_logfile1 /var/lib/mysql/ib_logfile1.bak
2. Start MySQL/MariaDB server:
   systemctl start mariadb

Also try to repair files using repair kit:
Log in to its interface by https://<IP>:8443/repair and using adminnistrator credentials
Then check file permissions and InnoDB files:
View attachment 28818

Contact submit a Plesk support for deeper investigation if the solution will not helped.

This does not work. Renaming the files and restarting simply results in the error that the files are unfound. This causes an exit. The service can not start.

 

[MHC_1]

If you don't have a full backup and get it to start with the above config, dump everything with `mysqldump --all-databases > backup.sql`, remove the innodb_force_recovery config, stop mariadb, run
```
mv /var/lib/mysql/ibdata1 /var/lib/mysql/ibdata1.bak
mv /var/lib/mysql/ib_logfile0 /var/lib/mysql/ib_logfile0.bak
mv /var/lib/mysql/ib_logfile1 /var/lib/mysql/ib_logfile1.bak
```

Restart mariadb and reimport your dump `mysql < backup.sql`

There is a way to check in the logs, which tables are corrupted and only dump those and reimport only those without rebuilding the whole database if you have a really big database but I won't get into it here

unfortunately

[root@server /]# mysqldump --all-databases > backup.sql
mysqldump: Got error: 1932: "Table 'apsc.aps_application' doesn't exist in engine" when using LOCK TABLES