SQL_THREAD Abort on BFA

[Jervin R]

If a PXC node acts as async slave, and commit deadlock occurs where the SQL_THREAD loses, it dies without being restarted. We should either have this covered via slave_transaction_retries or add another flag or automatically restart slave after the deadlock and winning applier thread has completed.

2016-03-18 11:32:36 15475 [Note] WSREP: trx conflict for key (1,FLAT8)e504963e c7dd9ee5: source: 8e7f5b88-e6f1-11e5-b8e9-9b1fc9c9344c version: 3 local: 1 state: MUST_ABORT flags: 1 conn_id: 4920882 trx_id: 281642595 seqnos (l: 1303798, g: 2167806, s: 2167804, d: 2167734, ts: 24937341962655885) <--X--> source: ad083d51-e1fc-11e5-bdb0-0b410c239327 version: 3 local: 0 state: APPLYING flags: 1 conn_id: 4586315 trx_id: 276912316 seqnos (l: 1303797, g: 2167805, s: 2167804, d: 2167734, ts: 24956588347501994)
2016-03-18 11:32:36 15475 [Warning] Slave SQL: Error in Xid_log_event: Commit could not be completed, 'Deadlock found when trying to get lock; try restarting transaction', Error_code: 1213
2016-03-18 11:32:36 15475 [Warning] Info table is not ready to be used. Table 'mysql.slave_relay_log_info' cannot be opened.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PXC-531