MySQL Bugs: #70829: Make replication thread message less alarming if master is different version

Bug #70829	Make replication thread message less alarming if master is different version
Submitted:	6 Nov 2013 9:17	Modified:	11 Nov 2013 8:29
Reporter:	Simon Mudd (OCA)	Email Updates:
Status:	Verified	Impact on me:	None
Category:	MySQL Server: Replication	Severity:	S3 (Non-critical)
Version:	5.6.14	OS:	Any
Assigned to:		CPU Architecture:	Any

Description:
While following the official upgrade procedure from 5.5 GA to 5.6 GA I upgrade the slaves first. On starting the I/O thread you see this message:

2013-11-06 09:40:42 10816 [Warning] Slave I/O: Notifying master by SET @master_binlog_checksum= @@global.binlog_checksum failed with error: Unknown system variable 'binlog_checksum', Error_code: 1193
2013-11-06 09:40:42 10816 [Warning] Slave I/O: Unknown system variable 'SERVER_UUID' on master. A probable cause is that the variable is not supported on the master (version: 5.5.23-log), even though it is on the slave (version: 5.6.14-log), Error_code: 1193

While these messages are warnings, there´s  a reference to an Error code and it looks quite alarming compared to normal replication startup.

How to repeat:
start a slave with 5.6 connected to a master still running 5.5

Suggested fix:
Make the slave recognise the master version and adapt accordingly. If the master does not understand the command that is not really surprising so just warn about the major version difference and do not pollute the logs with something that can be quite alarming if the reader does not understand what is happening.

2013-11-06 09:40:42 10816 [Warning] Slave I/O: Slave and master are running different major versions of MySQL (master version: 5.5.23-log, slave version: 5.6.14-log). This may not be advisable except while upgrading servers in a replication chain.

This alternative message hides known differences between versions and references to configuration variables which it's likely the master may not support.

More verbose logging could then provide information about the fact that the maser does not understand binlog_checksum or server_uuid variables but really that is just a cause first issue.

Hello Simon,

Thank you for the report.
Verified as described.

Thanks,
Umesh

I've just got the same messages on the error log connecting a fresh MySQL 5.6 slave server to a MySQL 5.5 that act as a master. It seems strange to have a warning message associated to an error code (Error_code: 1193). If it's not something that will impact the environment since it's a matter of version differences, it's a good case to be adjusted by developers.

2014-10-07 14:13:50 15356 [Note] Slave I/O thread: connected to master 'repl56@192.168.168.202:3306',replication started in log 'ndb5-bin.006822' at position 33878131
2014-10-07 14:13:50 15356 [Warning] Slave I/O: Notifying master by SET @master_binlog_checksum= @@global.binlog_checksum failed with error: Unknown system variable 'binlog_checksum', Error_code: 1193

WB