Bad MySQL Replication Bug on 4.1.21

I’m going to check to see if this has been found and reported but this is a bad gotcha in 4.1.21 (which for the most part is very stable).

If you have a large max_allowed_packet value on some of your slave boxes but a small value on the others then replication will silently fail on one half of your cluster.

The problem is that SHOW SLAVE STATUS shows zero seconds behind master and all other values indicated that replication is functioning correctly:

           Slave_IO_Running: Yes
          Slave_SQL_Running: Yes
      Seconds_Behind_Master: 0

… if you tail the log file it shows that replication is failing:

070723 19:11:44 [ERROR] Slave I/O thread: Failed reading log event, reconnecting to retry, log 'mysql-bin.000658' position 80177987
070723 19:11:44 [Note] Slave: connected to master '',replication resumed in log 'mysql-bin.000658' at position 80177987

%d bloggers like this: