I posted this last week to open questions and have not gotten a reply, so I am opening a ticket in the hopes someone can point me in the right direction. Any assistance so I can resolve this is greatly appreciated.

I have three servers. One master and two are slaves. I have a process that does an insert ignore the master and then does it again 5 seconds later to ensure the row has been written.

The servers are MariaDB 5.5.50 (CentOS 7.3) and 5.5.34 (CentOS 6.5).

A few weeks ago after making no changes what so ever, one of the slaves started logging the 1062 error(s) to the sql_errrors.log and I can not figure out why.

If I check the binary log on all three servers, I can find the insert ignore and the second one 5 seconds later on all three servers, so I know it is being executed everywhere.

I have diffed the server configs to make sure they are all the same.

Compared the mysql database between the servers for any differences.

I have logged in as the replication account and checked:

show variables like 'sql_error%';
show variables like '%mode%'; all settings return the same.

Can someone point me at anything else I can check to understand why the behavior for the one slave changed?