Details
-
Bug
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
5.5.40
-
None
-
None
-
Gentoo Linux, x64
Description
On (only) my production server, whenever I wait a while and restart MariaDB, it completely crashes and burns, e.g.:
141110 16:18:26 [Note] /usr/sbin/mysqld: Shutdown complete
|
141110 16:18:26 [ERROR] mysqld: Got error 'File size bigger than expected' when trying to use aria control file '/var/lib/mysql/aria_log_control'
|
141110 16:18:26 [ERROR] Plugin 'Aria' init function returned error.
|
141110 16:18:26 [ERROR] Plugin 'Aria' registration as a STORAGE ENGINE failed.
|
141110 16:18:26 InnoDB: The InnoDB memory heap is disabled
|
141110 16:18:26 InnoDB: Mutexes and rw_locks use GCC atomic builtins
|
141110 16:18:26 InnoDB: Compressed tables use zlib 1.2.8
|
141110 16:18:26 InnoDB: Initializing buffer pool, size = 1.0G
|
141110 16:18:26 InnoDB: Completed initialization of buffer pool
|
141110 16:18:26 InnoDB: highest supported file format is Barracuda.
|
141110 16:18:26 InnoDB: Waiting for the background threads to start
|
141110 16:18:27 Percona XtraDB (http://www.percona.com) 5.5.40-MariaDB-36.1 started; log sequence number 18387624831
|
141110 16:18:27 [ERROR] Aria engine is not enabled or did not start. The Aria engine must be enabled to continue as mysqld was configured with --with-aria-tmp-tables
|
141110 16:18:27 [ERROR] Aborting
|
141110 16:18:27 InnoDB: Starting shutdown...
|
141110 16:18:28 InnoDB: Shutdown completed; log sequence number 18387624831
|
141110 16:18:28 [Note] /usr/sbin/mysqld: Shutdown complete
|
Endlessly. Indeed:
$ ls -l /var/lib/mysql/aria*
|
-rw-rw---- 1 mysql mysql 40960 Nov 10 16:03 /var/lib/mysql/aria_log.00000001
|
-rw-rw---- 1 mysql mysql 2554299001 Nov 10 16:03 /var/lib/mysql/aria_log_control
|
It's large. I don't use ARIA, at least not on purpose. Apparently it was built with "--with-aria-tmp-tables" and I see no way to control that (at least via Gentoo Ebuild).
Each time this happens, I just sudo rm /var/lib/mysql/aria* and get on with my life. But sometimes it takes me a while to realize it's happened. And I'd prefer it just not happen.
Given, as mentioned, this only happens on my production server, I'd prefer to do as little destructive/interfering diagnostics as possible. Could someone guide me through the best steps to get the right information to diagnose this?