Not sure if we have a known bug as seems fixed in 2.4.7 but might be coincidence so wanted to report this:

Our 2.4.1 system suddenly rolled over the eventID in the database for no reason. As far as we can tell, it uses the max(eventid) + 1, but this is clearly not working as it went from 15,161,039 to 393 - here are the events in the event table, sorted by clock:

'15161039','2','3','200','1448276685','1','0',NULL,NULL,'0','0'
'393','3','4','1446589','1448276693','1','0',NULL,NULL,'0','578612635'

This makes no sense at all, nor do we understand how this continues to work as the source code seems to cache a few of these at a time, always MAX(eventiid)+1, but that should be 15,161,040 but it continues to work on the lower numbers one by one, which should be impossible.

We worry this is hiding some underlying problem that is not clear, and could happen with other IDs, and of course eventually we will have collisions, which then can't be fixed.

We saw this because we use this ID for integration into our ticket system and we had some DB reports that show us how fast we ACK things, average problem duration, etc. and they all started failing.

Checking this now, we see that it suddenly started working fine again using the max eventid at 2015-12-03 10:31:23 - when we upgraded the server binary to 2.4.7