I made a little progress in isolating this issue:
Through a process of editing the maria10 repo and just changing the
final digit I tested upgrading one release at a time (lol) by doing
this:
yum --enablerepo=el8 update MariaDB-server
10.4.12 works
10.4.13 works
10.4.14 works
10.4.15 works
10.4.16 works
10.4.17 works
10.4.18 works
10.4.19 works
10.4.20 works
Feb 17 11:49:06 localhost mysqld[999270]: 2026-02-17 11:49:06 0 [Note]
/usr/sbin/mysqld (mysqld 10.4.21-MariaDB) starting as process 999270 ...
Feb 17 11:49:06 localhost mysqld[999270]: 2026-02-17 11:49:06 0 [Note]
InnoDB: Using Linux native AIO
Feb 17 11:49:06 localhost mysqld[999270]: 2026-02-17 11:49:06 0 [Note]
InnoDB: Mutexes and rw_locks use GCC atomic builtins
Feb 17 11:49:06 localhost mysqld[999270]: 2026-02-17 11:49:06 0 [Note]
InnoDB: Uses event mutexes
Feb 17 11:49:06 localhost mysqld[999270]: 2026-02-17 11:49:06 0 [Note]
InnoDB: Compressed tables use zlib 1.2.11
Feb 17 11:49:06 localhost mysqld[999270]: 2026-02-17 11:49:06 0 [Note]
InnoDB: Number of pools: 1
Feb 17 11:49:06 localhost mysqld[999270]: 2026-02-17 11:49:06 0 [Note]
InnoDB: Using SSE2 crc32 instructions
Feb 17 11:49:06 localhost mysqld[999270]: 2026-02-17 11:49:06 0 [Note]
mysqld: O_TMPFILE is not supported on /tmp (disabling future attempts)
Feb 17 11:49:07 localhost mysqld[999270]: 2026-02-17 11:49:07 0 [Note]
InnoDB: Initializing buffer pool, total size = 128M, instances = 1,
chunk size = 128M
Feb 17 11:49:07 localhost mysqld[999270]: 2026-02-17 11:49:07 0 [Note]
InnoDB: Completed initialization of buffer pool
Feb 17 11:49:07 localhost mysqld[999270]: 2026-02-17 11:49:07 0 [Note]
InnoDB: If the mysqld execution user is authorized, page cleaner thread
priority can be changed. See the man page of setpriority().
Feb 17 11:49:07 localhost mysqld[999270]: 2026-02-17 11:49:07 0 [ERROR]
InnoDB: corrupted TRX_NO 62f900540078f867
It seems like whatever is going on begins happening at version 10.4.21
and further.
Anyone have any ideas?
On 2026-02-17 11:23, Gordan Bobic via discuss wrote:
On Tue, 17 Feb 2026 at 17:56, Andrew Weaver <[email protected]> wrote:
Alright so I was able to get a little further (thank you so much):
Server version: 10.4.11-MariaDB MariaDB Server
...
MariaDB [(none)]> set global innodb_fast_shutdown = 0;
Query OK, 0 rows affected (0.000 sec)
MariaDB [(none)]> quit
Bye
[yum.repos.d]# systemctl stop mariadb
I then uninstalled 10.4.11, put the repos back to the default and
installed 10.5.29-3 from RHEL9 distribution.
When I started mariadb it says this:
2026-02-17 10:48:42 0 [ERROR] InnoDB: corrupted TRX_ID
62f900540078f866
2026-02-17 10:48:42 0 [Note] InnoDB: Retry with
innodb_force_recovery=5
2026-02-17 10:48:42 0 [ERROR] InnoDB: Plugin initialization aborted
with error Data structure corruption
2026-02-17 10:48:42 0 [Note] InnoDB: Starting shutdown...
2026-02-17 10:48:42 0 [ERROR] Plugin 'InnoDB' registration as a
STORAGE ENGINE failed.
2026-02-17 10:48:42 0 [Note] Plugin 'FEEDBACK' is disabled.
2026-02-17 10:48:42 0 [ERROR] Unknown/unsupported storage engine:
InnoDB
Anything else needed to check in 10.4.11 land prior to upgrading?
So you got it working with 10.4, without force-recovery, did a full
clean shutdown, and 10.5 still choked on the redo log??
The only thing I can think of is that you started 10.4 with
force-recovery still on so it came up read-only and did nothing on
shutdown to purge the redo log.
Anything interesting in the log during 10.4 startup?
_______________________________________________
discuss mailing list -- [email protected]
To unsubscribe send an email to [email protected]
