Recovering from 'duplicate entry' error in replication setup

[Charles Pearson]

Hey guys,

I have a master-master MySQL replication setup between two web servers. We make use of a hardware load balancer in front of the two servers. Recently, some inconsistencies were being noticed between the two servers, and I realized that both servers were show a "duplicate entry" error when doing a 'show slave status\G;'.

I have found out how to get around this problem (adding slave-skip-errors = 1062 to the my.cnf file), however, I need to know how to move forward from where I am now to get the servers back in sync. I see that both servers are no longer advancing through the relay log file (when issuing a 'show slave status\G;', the Read_Master_Log_Pos continues to increment, but Relay_Log_Pos does not). Does this mean that the servers have stopped replication completely ever since it hit the error? And if this is the case, can I simply make my change to the my.cnf file, restart the MySQL services, and both servers will replay all un-played relay log file entries to get to the point where both are synced up again?

Thanks for the help in advance.