We have seen this in lower IOS versions where you have the stackmaster running a larger stack like 6 plus switches. What we saw when this happened was it would stop responding to telnet requests to get into the switch for periods of time . We never had any indication from users that it was affecting normal traffic flow . Your only solution will be to contact the TAC  or reloading the switch or stack will fix it  for a period of time.  It's probably a bug where it gets a slow memory leak where it does not release memory resources like it should.