We have loads of jobs that would cause damage if more than one copy was run. To prevent this we have the following code at the start of them:

  !if jobcnt('jobname,user.account') > 1 then
  !  tellop WARNING: Job is already running. I'm terminating.
  !  eoj
  !  endif

A couple of system problems have resulted in a large backlog of jobs waiting to run. One of our jobs is never getting past the above check — because by the time it moves from WAIT to EXEC, our scheduler has streamed the next copy which joins the WAIT queue.

One way I can see around this is to do the following instead (we have VEsoft's MPEX):

  !setjcw mpexfaststart 1
  !mpex
  %if jscount('jobname,user.account&EXEC') > 1 then
  %  tellop WARNING: Job is already running. I'm terminating.
  %  eoj
  %  endif
  %exit
  !setjcw mpexfaststart 0

But how do we terminate a job already running, if we don't have MPEX?

Tony Summers replies:

One method is to build a file at the start of each job and/or check for the file's existence in the same job.