jeff@ISI-VAXA.ARPA (Jeffery A. Cavallaro) (10/31/85)
Over a period of time, I have noticed the following problems with generic
batch queues within a cluster environment:
1. Significant skew among the CPU time-of-day clocks causes havoc cluster
wide. Jobs in a generic batch queue may be prematurely released for
execution to a specific queue. (sounds probable)
2. Cluster formation/reduction activity appears to have a weird effect
on the various job controllers on the CPUs. Running batch jobs get
blown away by the JOBCTLr,
Has anyone else noticed any of these situations? They may just be big
coincidences.
Jeff