We have a Master - Master replication setup, MySQL 5.6. Only one Master is used, the other is for backup and failover (we'll call that the slave). The binlog_format is set to ROW, autoincrement settings are made to avoid conflicts.
The problem: Slave is halted due to Duplicate key errors.
We debugged the cause to be that bulk deletes made by cron jobs on the Master did not run (completely?!?) on slave. We are talking about tens of thousands of records that are NOT deleted from the slave. We did not find errors in the MySQL error log.
That leads to unsyncronized replica and errors when inserts are made using PK that should have been deleted on the Slave.
The tables are MyISAM.
Any idea why the bulk delete doesn't propagate properly on the replica?