Paul,
Did you solve your problem?: Most of the "out of heap errors" you get are, as you have already addressed, memory exhaustion. How many jobs do you execute per day? How many jobs at any given moment are you running? do you utilize any adapter connections? These are the questions you ask to appropriately size the amount of memory you dedicate to your server.
Also, when this problem recurs, it's helpful to fail over to backup and then back again. This is a less painful option than doing a full master restart.
Hope this helps,
-Dwayne
