I would also defer to Stan on this on the technical details. Yet I have also some run-ins with runaway jobs. A "tactic" that has sometimes worked (short of shutting down the system) is to abort everyone else one by one.
My tact on the matter is there must be some resource that your program wants and someone else has it. Once you've got rid of "the other guy", the resource is released an your job goes on its merry way to completion, or to the abortjob you sent to it.
The reason I suggest "one by one" is that once you've aborted the "culprit" and your job quits, you won't have to abort remaining users.
----------
From: Michael Anderson[SMTP:[log in to unmask]]
Sent: Thursday, November 07, 1996 11:06 AM
To: [log in to unmask]
Subject: Re: Unabortable job
Greetings,
When the payroll can be run I am in trouble!
The mgr of our Payroll dept called me this morning with a problem. I have checked everything I
can think of. No locks in effect, IO has aborted, tried to abortjob....nothing works. The
terminal and printer in the payroll dept are hung. I hate to think that I have to shut the entire
system down, but hey! the payroll must be done.
I understand what Stan said about "critical mode" and I'm sure that is what going on here.
But can anybody explain this Strange stuff, mpex "SHOWPROC" shows me the PIN's, but KILL
complains "Process ! does not exist." Here's what I mean...
-------------------------------[SNIP]---------------------------------------
%showproc job=s1246
QPRI CPUTIME STATE JOBNUM PIN (PROGRAM) STEP
#s1246, mgr.payroll,pub (ldev 187)
B100 0:00.496 WAIT S1246 225 (JSMAIN.PUB.SYS)
?-277 0:00.000 READY S1246 188 (???)
%kill 188
Processing 188
Process ! does not exist. (CIERR 11004)
%kill 225
Processing 225
Process ! does not exist. (CIERR 11004)
%showprocc job=s1246
QPRI CPUTIME STATE JOBNUM PIN (PROGRAM) STEP
#s1246, mgr.payroll,pub (ldev 187)
B100 0:00.509 WAIT S1246 225 (JSMAIN.PUB.SYS)
?-277 0:00.000 READY S1246 188 (???)
%showtime
THU, NOV 7, 1996, 10:06 AM
%console 68
CONSOLE HAS BEEN SWITCHED FROM LDEV 20 TO LDEV 68 .
%abortjob #s1246
%showtime
THU, NOV 7, 1996, 10:06 AM
%abortio 187
No I/O to abort for device !. (CIWARN 3028)
%showtime
THU, NOV 7, 1996, 10:07 AM
%showproc job=s1246
QPRI CPUTIME STATE JOBNUM PIN (PROGRAM) STEP
#s1246, mgr.payroll,pub (ldev 187)
B100 0:00.545 WAIT S1246 225 (JSMAIN.PUB.SYS)
?-277 0:00.000 READY S1246 188 (???)
%showclock
SYSTEM TIME: THU, NOV 7, 1996, 10:08:23 AM
CURRENT TIME CORRECTION: 0 SECONDS
TIME ZONE: 6 HOURS 0 MINUTES WESTERN HEMISPHERE
%console 20
CONSOLE HAS BEEN SWITCHED FROM LDEV 68 TO LDEV 20 .
%
---------------------------[SNIP]-----------------------------------------
Has anyone every experienced this?
It seems as though MPE is a little out-to-lunch here. Showing me the PIN's and then telling me
they don't exist.
It would nice if somebody could shed some light on this. Seems like a RE-Boot is in order.
--
_______________________________________________
Michael Anderson
HP3000 Systems Programmer/Analyst
_______________________________________________
Current: TIW Corporation Houston Texas
Prior: MICOM Systems Simi Valey, Calif.
Email@work mailto:[log in to unmask]
Email@home mailto:[log in to unmask]
|
