[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] Problems with jobs



It is not possible unfortunately.

My submit machine is isolated and running Grid middleware (globus toolkit)
and on a completely different network. Im using the new GCB functionality
to get through a firewall into the cluster.

I have to get it running on this machine somehow

Chris
----- Original Message ----- From: "Ian Chesal" <ICHESAL@xxxxxxxxxx> To: "Chris Miles" <chrismiles@xxxxxxxxxxxxxxxx>; "Condor-Users Mail List" <condor-users@xxxxxxxxxxx>
Cc: <matthew.hope@xxxxxxxxx>
Sent: Thursday, December 08, 2005 5:27 PM
Subject: RE: [Condor-users] Problems with jobs


Is there anyway arround this condor_shadow problem or move them to
another
machine? Does it not defeat the point in having a powerfull 25 machine
cluster
here yet I need to end up with 50 processes running on my dainty
little
submitting
machine that it cant handle?? I would be better running my 1000 jobs
on
the
machine through a script in a loop :-s

None that I know of. Although that'd be neat feature: spawning shadows
on a dedicated "shadow" machine. I could actually use something like
that here at Altera.

I have now logged into my actual cluster.

These machines are all the same arch, memory etc.

2GB memory and 4Ghz CPUs - fairly powerfull.

I then submit a job... still not many of the vms being used up. Only 8
showing here
but there is upwards of 30 condor_shadow processes ? why not that many
busy machines?

It really sounds like your submit machine can't handle the load. What
happens if you move your submit node to one of your more power machines?

- Ian