Have you looked at SHOWPROC to monitor the CPU used?
I have some MPE CI scripts that do this, and I can send them to you,
only you would need to modify them to suite your own needs.
Interested, I'll describe how I used SHOWPROC, maybe it'll be helpful.
More than ten years ago, on MPE 6.5 or maybe is was 7.0, I had a similar
HUNG JOB problem. The backup to tape DDS3 would sometimes hang, you know
those pesky tapes. The backup was as automated as I could get it to be,
I used Stans "ONLINE" utility to auto mount the tape, and Telemons mail
program to send messages to a list of email addresses if the backup had
a problem, looking at STOREJCW and number of files stored and so on.
However, in the event of a hung tape, hung store.pub.sys process, their
wasn't much I could do.
Until I looked at SHOWPROC.
So I create a separate job, that the backup would stream, ;in= five
minutes after the backup started.
This job was coded to stay in a WHILE loop until the backup logged off,
and in the while loop it used SHOWPROC to monitor the CPU used by the
job streamed it. It would parse the output from SHOWPROC, use the CPU
numbers related to the STORE process, the loop is run once a minute, and
CPU number is compared with the last loop.
If the difference is less than a set minimum two times in a row, I send
out an email, and ABORTIO on the tape drive to get the backup moving
again, in my case the backup had some additional work to do after the
store process.
It's kinda weird parsing the SHOWPROC output, but it worked. If this
sounds like something you want to look at, let me know, I can send the
GETCPU script, and maybe the job that calls it.
Cheers,
Michael Anderson.
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
|