| Bug #36757 | SQL thread stop | ||
|---|---|---|---|
| Submitted: | 16 May 2008 15:57 | Modified: | 5 Feb 15:09 |
| Reporter: | Cyril SCETBON | ||
| Status: | Open | ||
| Category: | Server: ClusterRep | Severity: | S1 (Critical) |
| Version: | mysql-5.1-telco-6.3 | OS: | Linux (debian etch) |
| Assigned to: | Mats Kindahl | Target Version: | |
| Tags: | 5.1.27-ndb-6.3.17-telco, MySQL, cluster, replication | ||
| Triage: | Triaged: D2 (Serious) / R6 (Needs Assessment) / E6 (Needs Assessment) | ||
[16 May 2008 16:01]
Cyril SCETBON
change severity
[16 May 2008 16:19]
Hartmut Holzgraefe
Can you attach the mysqld and cluster logs spanning the time of the incident to the bug report?
[16 May 2008 16:35]
Cyril SCETBON
No error on clusterlog, just local checkpoints messages
[16 May 2008 16:36]
Cyril SCETBON
mysqld error log
Attachment: mysqld.err (application/octet-stream, text), 5.49 KiB.
[6 Jun 2008 10:25]
Cyril SCETBON
version upgraded. We still get the same error : It was not possible to update the positions of the relay log information: the slave may be in an inconsistent state. Stopped in ./mysqld-relay-bin.000014 position 44473744
[10 Jun 2008 16:24]
Cyril SCETBON
any idea ? It seems to be correlated with workload
[5 Feb 15:09]
Cyril SCETBON
We're still getting the same error but we've noticed that in the binary log of the master we have something like : # at posi # at posi+1 .... # at posi+n #datei server id ... ... #datei+n server id ... When SQL thread stops at posi it doesn't work anymore (even if we restart mysqld). But if we use "CHANGE MASTER" to jump to position posi+n it works until the next similar error.
[18 May 15:45]
Jonathan Miller
Per Martin S. Assigning to Mat's for comments
[18 May 15:47]
Jonathan Miller
Assigned to Mat's for comment per Martin S.
Description: SQL_THREAD stop cause of the following error : It was not possible to update the positions of the relay log information the slave may be in an inconsistent state. Stopped in ./mysqld-relay-bin.000003 position 225463952 We're hitting this error when we make changes to a disk table. When we start again the SQL_THREAD there isn't any error : start slave SQL_THREAD; mysql> desc spp_disk02 -> ; +-------+---------------------+------+-----+-------------------+-----------------------------+ | Field | Type | Null | Key | Default | Extra | +-------+---------------------+------+-----+-------------------+-----------------------------+ | id | bigint(20) unsigned | NO | PRI | NULL | auto_increment | | ise | varchar(54) | NO | UNI | NULL | | | vc01 | varchar(350) | YES | | NULL | | | vc02 | varchar(11) | NO | | -1 | | | vc03 | varchar(11) | NO | | -1 | | | ch04 | char(10) | NO | | | | | ch05 | char(10) | NO | | | | | ch06 | char(1) | NO | | | | | ch07 | char(3) | NO | | | | | ts | timestamp | NO | | CURRENT_TIMESTAMP | on update CURRENT_TIMESTAMP | +-------+---------------------+------+-----+-------------------+-----------------------------+ How to repeat: generate requests of type : SELECT * FROM spp_disk02 followed by UPDATE spp_disk02 SET vc01 = '360;20070504;999#6065;20070604;999#3655;20070704;999#3393;20070804;999#3370;20070904;999#3564;20071004;999#3317;20071104;999#3379;20071204;999#3354;20080104;999#3339;20080204;999#3290;20080304;999#3389;20080404;999' WHERE ise = 'ID-SPP-100-gciigJW4NGkLJElBlgFh23DhdkKZZIw1IYlaMIeZNAU'