The eventlogging_sync.sh script that is used to do custom replication of EventLogging tables from m4-master will not work with parallelized MySQL EventLogging consumer processes. There is a race condition in that script that causes the 'slave' database from not receiving all events. This has been happening at least since we started running parallel MySQL consumers back on Dec 3.
eventlogging_sync.sh needs something more reliable to sync on other than timestamps. We should bring back autoincrement ids on all EL tables, and then adapt eventlogging_sync.sh to use the id instead of a timestamp.