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

Re: [HTCondor-users] Remove mistyped Request...?

Steffen, condor_qedit will not automatically fix the Requirements _expression_ when you change the RequestGPUs to a RequestCPUs.  You have to use qedit to edit the requirements as well. 

For complex changes like requirements, it's best to put the changes into a file and then use  the -edits option of condor_qedit.

There are two ways to go about this.  One way would be to fix your submit file and then use

condor_submit -dump 

to have it spit out the correct job classad.  Grab the Requirements line and put it in a file. 

The other way is to use 

   condor_q  x.y -af:lrng  Requirements RequestGPUs > file

To make a file from job x.y and then edit the file.

Once you have your file of changes use 

  condor_qedit x.y -edits file

to make the changes.


From: HTCondor-users <htcondor-users-bounces@xxxxxxxxxxx> on behalf of Steffen Grunewald <steffen.grunewald@xxxxxxxxxx>
Sent: Thursday, April 22, 2021 4:47 AM
To: HTCondor Users Mailinglist <htcondor-users@xxxxxxxxxxx>
Subject: [HTCondor-users] Remove mistyped Request...?
Good morning,

a DAG recently submitted to my pool spawns jobs which accidentally set
RequestGPUs=1 - when RequestCPUs=1 was intended.

Apparently condor_qedit cannot fix this - RequestGPUs can be set to 0
(which is also reported by condor_q -better-analyze) but the comparison
"TARGET.GPUs >= RequestGPUs" still gets used for matching - while no
node provides a GPU, and TARGET.GPUs is undefined, causing the match
to fail.
Also the added RequestCPUs is not included into match criteria.

Is this a bug or a feature?


Steffen Grunewald, Cluster Administrator
Max Planck Institute for Gravitational Physics (Albert Einstein Institute)
Am Mühlenberg 1 * D-14476 Potsdam-Golm * Germany
Fon: +49-331-567 7274
Mail: steffen.grunewald(at)aei.mpg.de
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx with a
subject: Unsubscribe
You can also unsubscribe by visiting

The archives can be found at: