Hi, On Fri, Aug 12, 2016 at 12:26 PM, Kristian Nielsen <knielsen@knielsen-hq.org
wrote:
From a quick look at the patch, this seems to be because of code removed from innobase_kill_query(). Specifically:
@@ -4799,65 +5511,11 @@ innobase_kill_query( DBUG_ENTER("innobase_kill_query"); DBUG_ASSERT(hton == innodb_hton_ptr);
- if (!wsrep_thd_is_BF(trx->mysql_thd, FALSE) && - trx->abort_type == TRX_SERVER_ABORT) { - ut_ad(!lock_mutex_own()); - lock_mutex_enter();
Thanks for pointing this to me, I did have this code there some time ago but it also did not work out-of-the-box as underlying locking was changed. But now that I had fresh view on this I managed to solve mutexing problems (not exactly with above but kindof). However, this is not yet enough to get all rpl tests pass. There is still problem with two tests i.e. rpl_parallel and rpl_parallel_optimistic see: https://jira.mariadb.org/browse/MDEV-10550 in both cases slave refuses to sync with master. R: Jan