SUMMARY: consensus on rpc.lockd patch 100075 -again!

From: Maree Hegarty (maree@cs.uq.oz.au)
Date: Mon Jun 29 1992 - 20:34:08 CDT


Reply-to: Maree Hegarty <maree@cs.uq.oz.au>
Followup-to: Maree Hegarty <maree@cs.uq.oz.au>
This didn't make it out the first time as I was silly enough to
put "Re: SUMMARY" in the subject line...

Following my summary, I've received some more responses, which take us
a bit closer to consensus. We haven't installed -08 as yet, but will
in the near future (tomorrow!). Here is the previous summary and
what I've learnt since:

> I asked:
>
> > I'm wondering if anyone has come to a conclusion about rpc.lockd and
> > the 100075 patch revisions. Several people have commented here in
> > recent weeks. We too have tried earlier patch revisions
> > and consequently rebooted hosts due to hung processes. Is revision 08
> > safe?
>
> Doesn't sound like it...
>
> I only received three replies and no confident answers. Mixing revs
> of the patch is definitely a bad idea. rev 07 *seems* to work ok with
> 4.1.1, although one site was still experiencing occasional problems
> accessing WP or Lotus 123 files. Another site had only been running
> the patch for < 1 week. One site had tried to run both revs 07 and 08
> with 4.1.2 without success.
>
> Our own experience a few months ago with rev 07 and 4.1.1 was not a good
> one. We had lots of hung nntp processes.
>
> An earlier discussion in April indicated that the beta rev 08 caused
> problems but the official rev 08 was o.k. I'm not game to test it.
>
> If anyone has further comments, I'd appreciate them.
>
> Many thanks to:
>
> paul@hydres.uucp Paul Humphreys
> simon@liasun5.epfl.ch (Simon Leinen)
> Jim Ahokas ahokas@gun.com, nobugs!next8!ahokas@uunet.uu.net

Findings since this summary:

100075-08 is necessary for 4.1.2 or a mixture of 4.1.2 and 4.1.1
clients. Some sites have had success with rev 07 with 4.1.1 only
but I wouldn't recommend it. Rev 08 seems to be the way to go -
and use it everywhere.

A known problem with 100075-08 (and possibly all versions of) rpc.lockd
is locks on the server are not released when a client crashes if you use
TMPFS for /tmp. Sun apparently have a fix but only on special request.

Thanks to:

From: Christopher Hoover ch@csi.com
From: fallan@baobab.awadi.com.AU (Frank Allan - Network Manager)
From: millidc!blue!djm@uunet.uu.net (Drew Montag)
From: Dan Lorenzini <dal@gcm.com>
From: stiles@no2sun.cray.com (John Stiles)

--------------------------------------------------------------------
Maree Hegarty maree@cs.uq.oz.au Ph: +61 7 365 2864 Fax: +61 7 365 1999
Senior Systems Programmer, Computer Science, University of Queensland,
Queensland 4072, Australia



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:44 CDT