[Condor-users] Partial re-submission of cluster jobs

[George Graham <george.graham@xxxxxxxxxxxxxx>]

Hello,

 

I’ve got a question about re-submitting condor jobs if some jobs in a cluster fail.

 

I can re-submit my cluster of jobs with a submit file that contains:

+ID=$(Cluster)$(Process)

hold=true

 

then remove jobs I don’t want to re-run:

condor_rm –constraint ID!=5231.1

 

and then get the remaining job (5231.1) to run:

condor_release –all

 

If I want to re-run a number of jobs (with non-consecutive process numbers) is there away to combine logical expressions in the –constraint option of condor_rm? Ideally I’d like to be able to do something like this (which, of course, doesn’t work):

condor_rm -constraint  ID!=5231.1|ID!=5231.6| ID!=5231.15! |ID!=5231.50|ID!=5231.300

 

Does anyone have any ideas or workrounds?

 

Many thanks,

George