I had been having a problem with a couple of hundred uninterruptable
sleeping processes (it got up to 500+ in the end!!). From the responses I
got, it seems I just have to put up with them or reboot. We lost power in
the machine room this morning so my problem is fixed! I've reverted back
to the original ntalkd, so our users will have to live with flash bombs
for a while longer.
Thanks to
Dave Cherkus (cherkus_at_UniMaster.COM) who suggested doing:
Use 'dbx -k /vmunix' and:
(dbx) set $pid=10018 (from my example)
(dbx) trace
to find out why the processes are getting into such a state.
Paul Fardy (pdf_at_morgan.ucs.mun.ca) who said that talkd/ntalkd have been a
problem since V2.0 and had disabled it on his systems. As much as I'd
like to do this, I think our users may get a little upset. But then
again...
Julie (O'Shaughnessy) Battle (jab_at_wcsmail.com) who suggested killing the
parent process or debugging the ntalkd. I should have mentioned that the
parent process was init - but killing it did fix the problem!
Tom Webster (webster_at_i5142311.mdc.com) told me there is a bug in advfs
that can result in an unkillable process relating to a file-system timing
problem. This can be fixed by the patch OSF360-027. We don't do advfs,
but others may like to know this info anyway.
Thanks all,
Andrew
Andrew Moar : Ph (03) 9479 3928 email A.Moar_at_latrobe.edu.au
Systems Programmer, Information Technology Services
La Trobe University, Melbourne
Received on Fri Aug 02 1996 - 07:11:40 NZST