slave replication breaks after bug #74145 happens in master

[Jericho Rivera]

Originally filed upstream.

Description:
On a master-slave replication environment, following the steps to reproduce https://bugs.mysql.com/bug.php?id=74145 using FLUSH SLOW LOGS and not FLUSH LOGS. Slave replication breaks with:

Last_SQL_Errno: 0
Last_SQL_Error: Query caused different errors on master and slave. Error on master: message (format)='Invalid error code' error code=29 ; Error on slave: actual message='no error', error code=0. Default database: ''. Query: 'flush slow logs'

How to repeat:
Setup master-slave replication servers.

Do how-to-repeat steps in bug #74145 but use FLUSH SLOW LOGS instead of FLUSH LOGS.
[root@c7a mysql-5.6.32-linux-glibc2.5-x86_64]# chmod 000 data/c7a-slow.log
mysql> flush slow logs;
ERROR 29 (HY000): File './c7a-slow.log' not found (Errcode: 13 - Permission denied)

Check replication slave status.
Last_SQL_Error: Query caused different errors on master and slave. Error on master: message (format)='Invalid error code' error code=29 ; Error on slave: actual message='no error', error code=0. Default database: ''. Query: 'flush slow logs'

Suggested fix:
Same as FLUSH LOGS, avoid writing FLUSH SLOW LOGS to the binary log.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PS-1017