I am trying to compile Percona XtraDB Cluster on an OpenSolaris-based platform.
After I compiled it successfully (with some efforts…), I tried to build a 2-node multi-master and run sysbench oltp test, connecting through haproxy.
I found that sysbench always fails with ER_QUERY_INTERRUPTED error. After that I turned on wsrep_debug in my.cnf, and I found that it failed with
121120 14:19:39 [Note] WSREP: cluster conflict due to high priority abort for threads:121120 14:19:39 [Note] WSREP: Winning thread: THD: 1, mode: applier, state: executing, conflict: no conflict, seqno: 5727 SQL: (null)121120 14:19:39 [Note] WSREP: Victim thread: THD: 209, mode: local, state: executing, conflict: no conflict, seqno: 5726 SQL: UPDATE sbtest set k=k+1 where id=?121120 14:19:39 [Note] WSREP: BF kill (1, seqno: 5727), victim: (209) trx: 18714121120 14:19:39 [Note] WSREP: Aborting query: UPDATE sbtest set k=k+1 where id=?121120 14:19:39 [Note] WSREP: kill trx QUERY_EXEC for 18714121120 14:19:39 [Note] WSREP: kill query for: 209121120 14:19:39 [Note] WSREP: client rollback due to BF abort for (209), query: (null)121120 14:19:39 [Note] WSREP: Deadlock error for: (null)
It seems that it is sort of multi-master conflict (of course expected), am I correct? And is it normal to throw ER_QUERY_INTERRUPTED error in this case?