[MDEV-31544] MariaDB crash on OPTIMIZE with "MDL BF-BF conflict" Created: 2023-06-26  Updated: 2024-01-17

Status: Open
Project: MariaDB Server
Component/s: Galera
Affects Version/s: 10.9.5
Fix Version/s: None

Type: Bug Priority: Major
Reporter: William Edwards Assignee: Unassigned
Resolution: Unresolved Votes: 1
Labels: None


 Description   

When running OPTIMIZE, MariaDB crashes with "MDL BF-BF conflict". The node on which the OPTIMIZE runs is part of a Galera cluster.

This issue started occurring on 22-06-2023. MariaDB was last updated on 24-02-2023.

What changed is that a somewhat large database (1.2 GiB on disk) was added to the cluster.

The crash occurred while OPTIMIZE'ing different tables on both runs. These tables are part of the two largest databases, but are not the largest tables, relatively small even (36.0 MiB and 52.0 MiB respectively).

 Comments   
Comment by Alice Sherepa [ 2023-06-26 ]

Is it possible for you to try 10.9.6 (in case it is similar to MDEV-30303)? Also please add the error log.

Comment by William Edwards [ 2023-06-26 ]

I will upgrade when there is a release that includes MDEV-31367.

The error log was rotated.

Comment by William Edwards [ 2023-08-30 ]

Issue still occurs on 10.9.8.

Error log:

Aug 30 01:19:58 http-msh02 mariadbd[1172]: 2023-08-30  1:19:58 2196100 [Note] WSREP: MDL BF-BF conflict
 
Aug 30 01:19:58 http-msh02 mariadbd[1172]: schema:  <stripped>
 
Aug 30 01:19:58 http-msh02 mariadbd[1172]: request: (2196100 #011seqno 482884519 #011wsrep (toi, exec, aborted) cmd 3 45 #011OPTIMIZE TABLE `email_forwards`)
 
Aug 30 01:19:58 http-msh02 mariadbd[1172]: granted: (1 #011seqno 482884520 #011wsrep (high priority, exec, committing) cmd 0 161 #011(null))
 
Aug 30 01:19:58 http-msh02 mariadbd[1172]: 2023-08-30  1:19:58 2196100 [ERROR] Aborting

This issue occurred on a different database before; the largest database on the cluster. The database that the issue occurred on this time, is the second largest database on the cluster. The on-disk size difference is 5,12 GiB.

Comment by Rick Tuk [ 2024-01-17 ]

we are experiencing this issue on 10.6.16

  439 Jan 17 03:00:11 node02 mariadbd[83054]: 2024-01-17  3:00:11 9 [Note] WSREP: MDL BF-BF conflict
 
  440 Jan 17 03:00:11 node02 mariadbd[83054]: schema:  <stripped>
 
  441 Jan 17 03:00:11 node02 mariadbd[83054]: request: (9 #011seqno 76166763 #011wsrep (high priority, exec, executing) cmd 0 161 #011update `user` set `last_login` = '2024-01-17 02:00:10' where `id` = 'fda808be-8c27-46ac-b079-e79244      7ffe7e'ª4§e#023#001)
 
  442 Jan 17 03:00:11 node02 mariadbd[83054]: granted: (7 #011seqno 76166760 #011wsrep (toi, exec, committed) cmd 0 45 #011OPTIMIZE TABLE `log_history_daily`)
 
  443 Jan 17 03:00:11 node02 mariadbd[83054]: 2024-01-17  3:00:11 9 [ERROR] Aborting

Generated at Thu Feb 08 10:24:40 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.