Test suite "wait_for_slaves()" doesn't check replication is running

[Baron Schwartz]

I found that the test suite will hang indefinitely if replication is actually broken and wait_for_slaves() is called. This happens, for example, if the sandboxes are set with BINLOG_FORMAT=ROW during setup. The 12346 server will die like this:

               Last_SQL_Error: Could not execute Update_rows event on table test.t3; Can't find record in 't3', Error_code: 1032; handler error HA_ERR_END_OF_FILE; the event's master log mysql-bin.000001, end_log_pos 228033

To avoid this, we should make wait_for_slaves check if replication is running, and BAIL_OUT if not, so we stop as soon as possible after breakage, and can thus try to fix it.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PT-986