Mysql master<->master replication getting entirely messed up. What config do I need to tweak?

[dormando]

> For us, this is about high availability rather than load balancing. I 
> think we can handle this in our application layer by switching to 
> whichever tracker is online, and some script on each tracker machine to 
> watch for a tracker to be down and start up if appropriate.
> 

Should be fine as long as the other tracker is completely stopped. We do 
the same thing for HA reasons, but just do it a little closer to the DB 
itself (so you get the extra tracker capacity).

-Dormando