[MDEV-6199] mysql_upgrade causes errors after MariaDB 5.5 to 10.0 upgrade Created: 2014-05-02  Updated: 2015-04-03  Resolved: 2015-04-03

Status: Closed
Project: MariaDB Server
Component/s: Storage Engine - InnoDB
Affects Version/s: 10.0.10
Fix Version/s: N/A

Type: Bug Priority: Minor
Reporter: Arjen Lentz Assignee: Unassigned
Resolution: Not a Bug Votes: 0
Labels: upstream
Environment:

RHEL5,Linux


 Description   

Running mysql_upgrade after a 5.5 to 10.0 upgrade causes lots of notices in the errorlog (one per table), below is one example:

2014-05-02 14:29:34 2ba01e3bb940 InnoDB: Error: Table "mysql"."innodb_table_stats" not found.
 
2014-05-02 14:29:34 2ba01e3bb940 InnoDB: Error: Fetch of persistent statistics requested for table "db"."tbl" but the required system tables mysql.innodb_table_stats and mysql.innodb_index_stats are not present or have unexpected structure. Using transient stats instead.

Suggested fix:
I reckon that the mysql system tables should be upgraded first, before any other databases.

 Comments   
Comment by Arjen Lentz [ 2014-05-02 ]

Hmm this could be an observation fail on my part - InnoDB does report these, but when I upgraded another server it already chucks the warnings before mysql_upgrade is started. So it appears to be a consequence of the InnoDB storage engine starting up. Perhaps the loading of the buffer pool? Anyhow, it looks a tad ugly and perhaps something can be tweaked.

Comment by Elena Stepanova [ 2014-05-04 ]

I don't see the error appear on server startup (even with innodb_buffer_pool_load_at_startup=1 if that's the suspect). I do see them appear on opening tables, e.g. on a client connecting with a default database and with auto-rehash; I'm sure there are other ways why tables would be open and thus the error would occur.

Upon table opening, the errors happen in MySQL 5.6 as well, so if it's a bug, it's an upstream one.

However, the server error log upon upgrade before running mysql_upgrade is extremely ugly anyway, with all those "wrong structure" errors (understandably so), so I don't think additional ones from InnoDB make a big difference, especially since they tell the truth.

Comment by Elena Stepanova [ 2015-04-03 ]

Closing as not a bug for now, please comment to re-open if disagree.

Generated at Thu Feb 08 07:10:06 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.