[HTCondor-users] Reroute failed jobs

[Jin Mao <jin@xxxxxxxxxxxxxxxxxx>]

HTCondor community,

I am looking for a solution to reroute certain jobs which were held/released (JobRunCount>1) to run on a different set of servers. These set of servers will not serve any new jobs.

Existing job routing or job transform seems to apply only when the job was submitted to the cluster. Plus, JobRunCount only gets evaluated after the job is running for the first time.Â

Any suggestions? If classad of a job can be updated after this job is placed back in the queue or before it runs again, that may work too.

Thanks.

Jin.