Resolved Backups failing since a couple days: Unable to run the backup agent

[Talistech]

Server operating system version

Ubuntu 22.04.5 LTS

Plesk version and microupdate number

Plesk Obsidian 18.0.64 Update #1 Web Host Edition

Hi fellow Pleskians,

My backups recently started failing a couple days ago. This is the error I am getting:

<?xml version="1.0" encoding="UTF-8"?>
<execution-result status="error">
<message severity="error">
<description>Unable to run the backup agent: DB query failed: SQLSTATE[HY000] [2002] Connection refused

</description>
</message>
</execution-result>

Upon checking the journalctl logs while trying to take a backup, I see that it restarts every couple minutes because it crashes itself. You cna find the log trace on the bottom.
As far as I can see the most important parts are:
- [ERROR] InnoDB: Page old data size 7773 new data size 7772, page old max ins size 8457 new max ins size 8458
- 0x7fda800c9640 InnoDB: Assertion failure in file ./storage/innobase/btr/btr0cur.cc line 4262
- InnoDB: Failing assertion: page_zip || optim_err != DB_UNDERFLOW
- 241025 15:56:03 [ERROR] mysqld got signal 6 ;
- Query (0x7fda38010ad0): INSERT INTO `PleskStats` VALUES ('backup_type_full', 1) ON DUPLICATE KEY UPDATE `value` = `value` + 1

I have already done a full check of the database. There are no infected or corrupted tables.
I have also followed these guides:
- https://support.plesk.com/hc/en-us/...-MySQL-MariaDB-databases-on-Plesk-for-Windows
- How to fix InnoDB corruption cases for the MySQL/MariaDB databases on Plesk for Linux? - Support Cases from Plesk Knowledge Base
And I have tried to fix corrupted pieces by launching mysql/mariadb with innodb_force_recovery with ranges from 1-6.

Anyone has encountered this before or has any idea?
Thanks, much appreciated.
Mustafa

I was unable to paste the stack trace here so I uploaded it to pastebin.
This is the log trace: https://pastebin.com/raw/yehBNVn6

 

[Talistech]

Posting this for future reference;
Solved the issue. After all I still had a corrupt table.

I restarted the database with innodb_force_recovery = 4. After that I did a "plesk repair db" and I saw the following in my journalctl:

Oct 25 19:05:21 [REDACTED] mariadbd[2401316]: 2024-10-25 19:05:21 125 [ERROR] InnoDB: Summed data size 7810, returned by func 7811
Oct 25 19:05:21 [REDACTED] mariadbd[2401316]: 2024-10-25 19:05:21 125 [ERROR] InnoDB: Apparent corruption in space 124 page 5 of index `PRIMARY` of table `psa`.`PleskStats`
Oct 25 19:05:21 [REDACTED] mariadbd[2401316]: 2024-10-25 19:05:21 125 [ERROR] InnoDB: In page 5 of index `PRIMARY` of table `psa`.`PleskStats`

I went into the database, exported the table `PleskStats`. Dropped the table `PleskStats` and reimported the table. It all works fine now again.

Good evening everybody.