create...select processing needs to be verified

Bug #904172 reported by Seppo Jaakola
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
MySQL patches by Codership
Fix Released
Medium
Seppo Jaakola
5.1
Fix Released
Medium
Seppo Jaakola
5.5
Fix Released
Medium
Seppo Jaakola

Bug Description

Currently create..select is not run under total order isolation, at least with ROW binlog format.
It should be verified that this does not cause issues both with ROW and STATEMENT formats.

Also, create..select with temporary table can crash the master node. This seems to happen when select contains join condition, here's a test for reproducing:

CREATE TABLE t1 (i int);
INSERT INTO t1 values (1),(3);
CREATE TABLE t2 (j int);
INSERT INTO t2 values (2),(4);
CREATE TEMPORARY TABLE tt as select t1.i, t2.j from t1, t2;

As a result, following assert will fire:

111214 11:51:37 [Note] WSREP: empty rbr buffer, query: insert into t1 values (3,3)
mysqld: galera/src/replicator_smm.cpp:812: virtual wsrep_status_t galera::ReplicatorSMM::post_commit(galera::TrxHan
dle*): Assertion `trx->state() == TrxHandle::S_COMMITTING || trx->state() == TrxHandle::S_REPLAYING' failed.
111214 11:51:37 - mysqld got signal 6 ;

Revision history for this message
Seppo Jaakola (seppo-jaakola) wrote :
Revision history for this message
Seppo Jaakola (seppo-jaakola) wrote :
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.