[Condor-users] JobPrio not being obeyed on a schedd?

["Ian Chesal" <ICHESAL@xxxxxxxxxx>]

Title: JobPrio not being obeyed on a schedd?

I have a user with two clusters of jobs. One cluster has the JobPrio set to 10 on all the processes. The other has the JobPrio set to 12 on all the processes. They are, for all purposes, the exact same cluster of jobs. Same expected memory, disk, etc. And the two clusters are competing for the same few machines.

Watching these two clusters run over the last 14 hours the cluster with JobPrio == 12 seems to be running a few more jobs than the cluster with JobPrio == 10. But, why is the cluster with JobPrio == 10 even running at all? I thought the job priorities determined the absolute execution order for a single users jobs on a single schedd. Am I mistaken? How can I figure out why the schedd thought it would be a good idea to pass off lower priority jobs instead of higher priority jobs to the startds?

- Ian