Bug #41140 | Failure of replication on triggered insert if innodb_autoinc_lock_mode > 0 | ||
---|---|---|---|
Submitted: | 30 Nov 2008 23:41 | Modified: | 28 Aug 2009 16:40 |
Reporter: | Elena Stepanova | Email Updates: | |
Status: | Can't repeat | Impact on me: | |
Category: | MySQL Server: Locking | Severity: | S2 (Serious) |
Version: | 5.1.22-rc, 5.1.30 | OS: | Any |
Assigned to: | Andrei Elkin | CPU Architecture: | Any |
[30 Nov 2008 23:41]
Elena Stepanova
[7 Jan 2009 10:35]
Borus Uchitel
I've find that the current value of the autoinc column is not advancing in the slave under the conditions defined in the description. In my case, when failing over to the slave server, the inserts in the table produced rows with lower autoinc values then expected.
[23 Jun 2009 19:14]
Sveta Smirnova
Bug #45677 about same behavior, but with innodb_autoinc_lock_mode=0
[25 Jun 2009 9:25]
Magne Mæhre
Jon Olav and I have looked at this one, and we're not able to reproduce this one on neither Linux nor Solaris. The similar bug #45677 is, on the other hand, reproducible. I'm unassigning myself, and recommend that the replication team checks both of these.
[29 Jun 2009 12:06]
Elena Stepanova
I agree -- I still can reproduce the problem (replication failure) as described on versions 5.1.22-5.1.30, but not on 5.1.31 or higher. Regarding bug#45677, see also bug#26316 which might be related.
[28 Aug 2009 16:40]
Andrei Elkin
Could not reproduce the problem with the current built. There must have been some changes in the code done. Even the preparation step show create table l reports CREATE TABLE `l` ... ENGINE=InnoDB AUTO_INCREMENT=3 i.e 3 not the claimed 4.