random failures in flush_block_commit test

Bug #673988 reported by Andrew Hutchings
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Drizzle
Opinion
High
Andrew Hutchings
7.0
Opinion
High
Andrew Hutchings

Bug Description

main.flush_block_commit [ fail ]
drizzletest: At line 49: query 'reap' failed: 1205: Lock wait timeout exceeded; try restarting transaction

The result from queries just before the failure was:
drop table if exists t1;
create table t1 (a int) engine=innodb;
begin;
insert into t1 values(1);
flush tables with read lock;
select * from t1;
a
commit;
select * from t1;
a
unlock tables;
begin;
select * from t1 for update;
a
1
begin;
select * from t1 for update;
flush tables with read lock;
commit;

Related branches

Changed in drizzle:
importance: Undecided → Critical
assignee: nobody → Andrew Hutchings (linuxjedi)
milestone: none → 2010-11-22
Revision history for this message
Lee Bieber (kalebral-deactivatedaccount) wrote :

Andrew, we are seeing this again. we are only seeing this on our new OSX machines. The only difference I can see on these machines is that they are running boost 1.44, none of our other machines are running boost 1.44
-

Revision history for this message
Andrew Hutchings (linuxjedi) wrote :

Also started happening on Fedora 12 again :/

Revision history for this message
Andrew Hutchings (linuxjedi) wrote :

Can't reproduce on the Intel F12 box. Must be something unique about slicehost-fedora12-64
 and the OSX machines. I doubt it is boost 1.44 or I would see it locally.

Revision history for this message
Andrew Hutchings (linuxjedi) wrote :

Other boxes have now failed with this too.

Revision history for this message
Andrew Hutchings (linuxjedi) wrote :

after 3000 runs of this test during high load situations on one of the affected boxes I'm pretty certain its not load related. It must be triggered by a previous test.

Revision history for this message
Andrew Hutchings (linuxjedi) wrote :

Still unable to reproduce this. Keep hitting bug#680028 when attempting variants on this

Revision history for this message
Andrew Hutchings (linuxjedi) wrote :

This hasn't failed in over a week. Setting to Opinion for now as I believe some of the locking changes recently as well as InnoDB upgrades have fixed it.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.