[MDEV-769] LP:884715 - Aria spontaneous corruption Created: 2011-11-01  Updated: 2014-03-18  Resolved: 2014-03-18

Status: Closed
Project: MariaDB Server
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: Bug Priority: Major
Reporter: Dreas van Donselaar (Inactive) Assignee: Unassigned
Resolution: Incomplete Votes: 0
Labels: Launchpad

Attachments: XML File LPexportBug884715.xml    

 Description   

We've duplicate an Aria table to some servers by copying the files directly. A mysqlcheck confirms that the table is good, aria_chk also sees no issues. After a restart of MariaDB all is fine with the table as well. Now we checked again the next day, and suddenly a check indicates the table is corrupted. This happened twice now (on the same server).

:~/$ mysqlcheck -v database table
 
database.table
 
warning  : Table is marked as crashed
 
error    : Wrong base information on indexpage at page: 34022
 
error    : Corrupt
 
:~/$

The actual count of rows using {{select count(*)} did not change. However, after executing a repair table many rows disappeared. Nothing is writing to this table (only reads in very low volume), so it's really strange that this spontaneously corrupts without any restarts/writing to the table.

I'll upload the original non-crashed version, and the crashed-version to FTP.

 Comments   
Comment by Elena Stepanova [ 2012-06-07 ]

Re: Aria spontaneous corruption
Hi Dreas,

There have been quite a lot of changes, including fixes for related or similar bugs you filed; so it's quite possible that this one got fixed along with them. Could you please let us know if you still experience the problem?

Thank you.

Comment by Dreas van Donselaar (Inactive) [ 2012-06-07 ]

Re: Aria spontaneous corruption
It has not recently reoccurred.

Comment by Rasmus Johansson (Inactive) [ 2012-06-07 ]

Launchpad bug id: 884715

Comment by roberto spadim [ 2013-06-19 ]

maybe there's some information in aria log? i don't know if i'm right, but aria have a file where it write changes and after writing to this file, it will write to the table file, like a ext3 filesystem with journal
maybe the lost rows are inside this journal file, i done something like this in past and to solve my problem, i first flushed tables and locked for read, after this i copied some files from one server to another, and i had no errors

Comment by Elena Stepanova [ 2014-03-18 ]

It has never contained enough information to reproduce the issue, and later the reporter said it stopped happening. Closing.

Generated at Thu Feb 08 06:31:12 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.