Mailing List Archives
Authenticated access
|
|
|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] Why does machine reject job for unknown reasons
- Date: Tue, 15 May 2007 11:56:36 -0500
- From: Daniel Forrest <forrest@xxxxxxxxxxxxx>
- Subject: Re: [Condor-users] Why does machine reject job for unknown reasons
On Tue, May 15, 2007 at 05:02:11PM +0100, Alexander Dietz wrote:
>
> On 5/15/07, Ian Chesal <ian.chesal@xxxxxxxxx> wrote:
> >
> > Sorry. That's my fault. "is undefined" and the math stuff won't
> > work in a constraint tag. Although Steffen pointed out that you
> > mentioned your pool was 152 big so that means you're matching just
> > about every machine. Still, you can try:
> >
> > condor_status -const '( Arch == "X86_64" ) && ( OpSys == "LINUX" )
> > && ( CkptArch == "X86_64" ) && ( CkptOpSys == "LINUX" )
> > && ( Disk >= 10000 ) && ( Memory >= 9 )'
> >
To clear up some misconceptions here, math does indeed work in a
constraint expression, but "is undefined" appears to be how condor_q
-better-analyze prints the expression '(Value =?= UNDEFINED)' and is
not a valid construct. The expression you want is:
condor_status -const '( Arch == "X86_64" ) && ( OpSys == "LINUX" ) && ( ( CkptArch == "X86_64" ) || ( CkptArch =?= UNDEFINED ) ) && ( ( CkptOpSys == "LINUX" ) || ( CkptOpSys =?= UNDEFINED ) ) && ( Disk >= 10000 ) && ( ( Memory * 1024 ) >= 10000 )'
Note that "CkptArch" and "CkptOpSys" both come from the JobAd, so the
tests for "UNDEFINED" are needed to get any output from condor_status.
An equivalent expression would be:
condor_status -const '( Arch == "X86_64" ) && ( OpSys == "LINUX" ) && ( Disk >= 10000 ) && ( ( Memory * 1024 ) >= 10000 )'
Which is almost what you used here:
> Still no output. The command
>
> condor_status -const '( Arch == "X86_64" ) && ( OpSys == "LINUX" )
> && ( Disk >= 100 ) && ( ( Memory * 1024 ) >= 100 )'
>
> returns 150 machines... Maybe it is related to checkpointing? Maybe
> better to try the vanilla universe instead of standard?
That is a possibility. What version of Condor are you running? I
believe you need at least 6.8.2 to support checkpointing on X86_64.
I would certainly start with getting a simple shell script to run as a
Vanilla job before trying anything more complex.
--
Daniel K. Forrest Laboratory for Molecular and
forrest@xxxxxxxxxxxxx Computational Genomics
(608) 262 - 9479 University of Wisconsin, Madison