Bug #39502 | myisam table crash on update or replace into periodically | ||
---|---|---|---|
Submitted: | 17 Sep 2008 17:21 | Modified: | 1 Nov 2008 4:43 |
Reporter: | Steve Drew | Email Updates: | |
Status: | No Feedback | Impact on me: | |
Category: | MySQL Server: MyISAM storage engine | Severity: | S2 (Serious) |
Version: | 4.1.22+4.1.20 | OS: | Linux (Centos 4.7) |
Assigned to: | CPU Architecture: | Any |
[17 Sep 2008 17:21]
Steve Drew
[17 Sep 2008 17:23]
Steve Drew
As a note, when the slaves have crashed, the position of the wrong key is different across all the slaves.
[18 Sep 2008 17:15]
Sveta Smirnova
Thank you for the report. Have you tried to find any pattern after which error occurs? Like particular DELETE query in bug #22384?
[18 Sep 2008 17:32]
Steve Drew
REPLACE INTO are the main culprits (95%), with the very occasional UPDATE causing problems. I encountered the problem 11 hours after repairing the master and slaves servers last night. When the issue occurs, its appearing that it affects a small range (around 80 items), based around the object_name (first field).. And it records that have been inserted today where the problems occur.
[19 Sep 2008 6:01]
Steve Drew
MyISAM file: db/probtable Record format: Packed Character set: latin1_swedish_ci (8) File-version: 1 Creation time: 2008-02-04 16:30:59 Recover time: 2008-09-17 10:02:02 Status: changed Data records: 5727548 Deleted blocks: 0 Datafile parts: 5728665 Deleted data: 0 Datafile pointer (bytes): 4 Keyfile pointer (bytes): 4 Datafile length: 3272684764 Keyfile length: 4295779328 Max datafile length: 4294967294 Max keyfile length: 4398046510079 Recordlength: 1519 I think I've found the cause of this. If the keyfile length is > Max datafile length and the thread_concurrency variable is set, then corruption occurs extremely quickly. I'll check to see if this is a know bug already reported and update accordingly. This does not appear to be an issue if the slave is running 5.0.22
[19 Sep 2008 9:33]
MySQL Verification Team
probably a duplicate of bug #31305
[1 Oct 2008 4:43]
Valeriy Kravchuk
Please, upgrade to 4.1.24 and check if the problem is repeatable in this newer version.
[2 Nov 2008 0:00]
Bugs System
No feedback was provided for this bug for over a month, so it is being suspended automatically. If you are able to provide the information that was originally requested, please do so and change the status of the bug back to "Open".