Hello,
I am sure this info is somewhere on the web but have not been able to find it via google or by searching this site. On HPUX we have a model script that runs a command to capture the spool file to a repository and doesn't actually physically print the spool file. This has been working fine for many years.
Today we found a strange quirk. If the print queue is disabled, the model script stops executing as expected but if the capture command is running, it continues to run until it completes. The return code from the capture command is not passed back to the model script (or the model script ignores it because it has been disabled). When the queue is enabled again, the job that was running when the queue was disabled is processed again. The lpsched man page (lpsched 1M on HP) mentions this behavior and this is the behavior I would expect from lpsched. The problem is the same spool file is being completely processed twice causing duplication in our repository.
So here comes the question. Does anyone know anyway I can send SIGTERM/SIGKILL/etc to the capture command when the queue is disabled by placing code in the same model script? In my case it would be approriate to kill the capture command if the queue is disabled. If this is not possible, how about the inverse - passing the success of the capture command back to model script thus completing the job so lpsched starts with the next job when the queue is enabled once again.
During some of my testing to figure this out, I have noticed that once a queue is disabled, any commands running from that model script seem to lose STDOUT and STDERR. Can anyone confirm this? I am doing more testing to see if I can prove otherwise.

Thanks,
TioTony

I think that you need to smarten up this capture command. It should simply decline to recapture a file that already exists. Or when it completes, it should see who its parent is. If the parent is init, it should undo what it just did.

Thanks Perderabo but the capture command is also used outside of the model script. Making that sort of change could affect other uses of the capture command. Maybe we can add a switch to it or something. That will take a while to change. Any other suggestions how I could handle this from within the script? Anyone know where I can find more info on how lpsched communicates with the model script?

Then change the model script to call a wrapper script. The wrapper script will run your capture program. When the capture program fishishes, the wrapper will look to see who its parent is. If it is still the model script, it just returns the return code from the capture program. But if it is init, it takes some special action. Maybe it removes the file that capture program just posted. Or maybe it cancels the current print job. Or maybe it searches the spool directory and removes all traces of the job. You need to bang on it until it starts working.

Thanks again.