I'm using OSG connect to analyze around 100 neural signals (one per process). Each process should take roughly the same amount of time to run, but I see that while most processes complete in <10 minutes, some are taking more than 50 minutes. I'm guessing this is because whatever server that job got sent to is just slower.
How should I be addressing this issue? Is there a command like `condor_release` to move processes that are taking a while to another server (not held jobs)?