I am having the same problem with Condor 6.6.10 on Fedora Core 5. Is there a way of allotting enough SWAP to Condor to prevent the limit being reached. I am told it is impossible to configure SWAP from the local_config only to set a SWAP reserve, but my Fedora Core does not allot any virtual memory at all.
I have set RESERVED_SWAP to O but still receive the message that there is not enough SWAP in the SCHEDD log.
I am investigating DOS commands that manage the Virtual Memory and would greatly appreciate advice.
From: Erik Paulson <epaulson@xxxxxxxxxxx>
Reply-To: Condor-Users Mail List <condor-users@xxxxxxxxxxx>
To: Condor-Users Mail List <condor-users@xxxxxxxxxxx>
Subject: Re: [Condor-users] swap space problem
Date: Fri, 2 Jun 2006 11:13:21 -0500
>On Fri, Jun 02, 2006 at 09:07:33PM +0530, P. Nagaraj wrote:
> > Hi,
> > I apologize I sent this same mail few minutes earlier mail with no
> > subject line.
> > I am running Condor Version 6.6.5. After upgrading the OS of the master
> > node to Linux with 2.6.9-22.0.1.ELsmp (ScientificLinux 4.2) , I find this swap
> > space problem in the Sched log of the condor master.
> > 6/2 19:07:39 Tables are consistent
> > 6/2 19:07:39 Swap space estimate reached! No more jobs can be run!
> > 6/2 19:07:39 Solution: get more swap space, or set RESERVED_SWAP = 0
> > 6/2 19:07:39 0 jobs matched, 13 jobs idle
>Condor 6.6.x has problems with memory detection on 2.6 kernels.
>Upgrade to Condor 6.7, or set RESERVED_SWAP = 0.
>Condor-users mailing list