Re: [HTCondor-users] basic configuration question #2

[Matthew Farrellee <matt@xxxxxxxxxx>]

On 06/18/2013 08:26 PM, Hugh Jennings wrote:
> Hi,
>
> I have installed HT Condor on a windows machine to access a 2 system
> test cluster.  If I execute condor_status from the windows machine I get
> a predictable result
>
> H:\condor\bin>condor_status
> Name               OpSys      Arch   State     Activity LoadAv Mem
> ActvtyTime
>
> slot1@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.240  493
>   0+00:00:04
> slot2@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.000  493
>   0+00:00:01
> slot3@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.000  493
>   0+00:00:02
> slot4@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.000  493
>   0+00:00:03
> slot5@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.000  493
>   0+00:00:04
> slot6@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.000  493
>   0+00:00:05
> slot7@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.000  493
>   0+00:00:06
> slot8@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.000  493
>   0+00:00:02
> slot1@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.280 6039
>   0+00:00:03
> slot2@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.000 6039
>   0+00:00:04
> slot3@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.000 6039
>   0+00:00:05
> slot4@xxxxxxxxxxxx LINUX      X86_64 Unclaimed Idle      0.000 6039
>   0+00:00:06
>                       Total Owner Claimed Unclaimed Matched Preempting
> Backfill
>
>          X86_64/LINUX    12     0       0        12       0          0
>       0
>
>                 Total    12     0       0        12       0          0
>       0
>
> but if I try to submit a simple job file such as:
>
> executable     = /bin/ls
> universe       = vanilla
> input          = /proc/cpuinfo
> output         = loop.out
> error          = loop.error
> log            = loop.log
> queue 150
>
>
> I get an error message
>
> H:\condor\bin>condor_submit ..\..\junk\simple_submit.txt
> Submitting job(s)
> ERROR: Executable file /bin/ls does not exist
>
> H:\condor\bin>
>
> During the installation on Windows I chose the "Do not run jobs on this
> machine" option.  I was expecting I could just submit jobs to the
> cluster. Why is it looking for a version of the binary on the local (I
> assume) Windows machine?
>
> Regards,
>
> Hugh

condor_submit is trying to be helpful, but not smart. It can't tell that 
/bin/ls should exist on the eventual execute machine. It's checking for 
it locally (windows) and failing.

You'll need something like "transfer_executable = false" to make 
condor_submit less helpful.

Next up it'll likely complain about the input file being missing, same 
reason. You'll want to look at should_transfer_files at that point, see 
man condor_submit.

Finally, condor_submit is going to be helpful and set a default 
Requirements expression for you that will only let the jobs run on 
windows. It assumes you are submitting from windows therefore you want 
to run on windows. You'll have to do something like requirements = OsSys 
=!= UNDEFINED to get around that issue.

Best,


matt