another possibility (depending upon the way in which you have your
accounting structure designed) could be the effect of any logon UDCs.
Without issuing anything except the !JOB and !EOJ in your trial, a logon
UDC for that user (ie: do you utilize a BA user with a catalog that differs
from the IA users?) could be trying to do all sorts of other things,
including executing other processes.   I've seen these hang a job literally
forever on a wait for lock or wait for data where either a deadlock has
occurred or the datasource is asleep.  Do you do anything, perhaps, with
message files, and your jobs are simply waiting for a feed from a semi-
asleep-at-the-switch IPC?
Since this is evidently sporadic, I'd be inclined to sleuth in that
direction, rather than the process queues/priorities.

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *