It would also be great to have a replication test to verify bug #1008278 is indeed fixed by this patch. It's not clear why only Master-Master replication was affected, could it be that SQL thread always crashes with userstat enabled? In both cases a test case should be rather trivial.
Vlad,
Please recommit and provide more descriptive revision comments, mentioning all three bug numbers, and the problem description. Here's an example of a good revision comment: http:// bazaar. launchpad. net/~laurynas- biveinis/ percona- server/ bug951588- 5.1/revision/ 447
It would also be great to have a replication test to verify bug #1008278 is indeed fixed by this patch. It's not clear why only Master-Master replication was affected, could it be that SQL thread always crashes with userstat enabled? In both cases a test case should be rather trivial.