[HTCondor-users] Startd/Collector refresh delay

[Jerome Samson <jeromes@xxxxxx>]

Hi,

I have a config issue with our startd service.
We setup the pool so that a condor job can be preempted if the 
NonCondorLoadAverage on the machine is to high.
So far it works fine but when a non-condor process taking much cpu ends, 
the loadaverage indicated for the machine does not falls back very 
quickly to its actual level.
There must be a sort of refresh delay which stucks the machine in a 
"busy" state when it could be starting new jobs...

I am not sure how to adress this:
Should i just change the MASTER_UPDATE_INTERVAL to a shorter value like 
60 (default is 300) or will this generate too much traffic ?

Any other idea ?

Best,

Jerome
+33 (0) 1 53 06 25 37
---------------------
BUF Compagnie
139-141 Boulevard Ney
75018 Paris
---------------------
Pensez a l'environnement avant d'imprimer ce message