[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] jobs match, but prefer another specific job despite its worse user-priority
- Date: Tue, 22 Jan 2008 18:50:10 +0100
- From: Harald van Pee <pee@xxxxxxxxxxxxxxxxx>
- Subject: Re: [Condor-users] jobs match, but prefer another specific job despite its worse user-priority
I have found out, that a condor_reconfigure does not help.
After a condor_restart everything seems to work as expected!
This behaviour is a little bit misleading. For example its possible
that a logfile reports the new value, but the service is still using the old
This I have explicitly seen for the negotiator, where I have changed
NEGOTIATOR_INTERVAL to 2 minutes, but obviously the old 5 minutes are still
1/22 15:59:30 ---------- Finished Negotiation Cycle ----------
1/22 16:00:06 Got SIGHUP. Re-reading config files.
1/22 16:00:06 NEGOTIATOR_INTERVAL = 120 sec
1/22 16:04:30 ---------- Started Negotiation Cycle ----------
On Thursday 17 January 2008 02:03 pm, Harald van Pee wrote:
> If I understand this correct, than a
> CLAIM_WORKLIFE = 120
> will keep a claim reserved for 120s if a job finishes faster the claim is
> still open and the user can start another job.
> But here we still see that a user with the lowest priority can use a couple
> of machines with 16 virtuell machines for hours even with jobs running
> longer than 10 minutes
> and CLAIM_WORKLIFE = 120 set
> and high priority user can not get any of them.
> Can someone explain such a behaviour?
Harald van Pee
Helmholtz-Institut fuer Strahlen- und Kernphysik der Universitaet Bonn