Date: Sun, 8 Nov 1998 11:35:53 -0700 (MST) From: David G Andersen <danderse@cs.utah.edu> To: vanmaren@fast.cs.utah.edu (Kevin Van Maren) Cc: bgrayson@marvin.ece.utexas.edu, freebsd-smp@FreeBSD.ORG Subject: Re: disk-wait problems/hangs Message-ID: <199811081835.LAA16723@lal.cs.utah.edu> In-Reply-To: <199811080638.XAA21120@fast.cs.utah.edu> from "Kevin Van Maren" at Nov 7, 98 11:38:48 pm
next in thread | previous in thread | raw e-mail | index | archive | help
We've been experiencing the same problems; on this end, it looked like the
problem was possibly related to amd/nfs. We're running in a NIS
environment with heavy AMD and NFS usage; disabling things like SMP didn't
solve the problem, but disabling AMD did. Our solution was to back AMD
out to the pre-aug-23 integration of the a16 version; another person has
upgraded theirs to the latest b1 release. (We tried the upgrade, but it
didn't solve the problem).
We're trying to get a crashdump of the hang, but after the AMD backout,
it's become difficult to reproduce.
Interestingly enough, the a.out netscape is _exactly_ how we reproduce the
problem. :) We don't see things stuck in 'D', but it may be that they're
going downhill so quickly that the entire machine hangs before we catch
it.
(I'm not on -smp, so please CC: responses to me; a colleague forwarded me
the message).
-Dave
Lo and behold, Kevin Van Maren once said:
>
> > From: "Brian C. Grayson" <bgrayson@marvin.ece.utexas.edu>
> > Date: Sat, 7 Nov 1998 23:36:40 -0600
> > To: freebsd-smp@FreeBSD.ORG
> > Subject: disk-wait problems/hangs
> >
> > We're running 3.0-RELEASE on a few dual P-II boxes.
> > Occasionally, processes will start getting hung in 'D'
> > (disk-wait, IIRC), even on an otherwise-idle machine. They
> > never come out, they aren't kill -9'able. Once the system gets
> > into this state, commands like 'df' and 'ls' are likely to go into
> > disk-wait. Eventually (on the order of minutes/hours),
> > something crucial like nfsd, ypbind, or sshd gets stuck in D,
> > and the machine requires a reboot.
> >
> > I can reproducibly force the cascade of D problems by running
> > an a.out Netscape -- it gets hung after <2 CPU seconds, and
> > things go downhill quickly. But I believe the problems have
> > occurred before without the use of any a.out executables.
> >
> > Has anyone else seen this?
> >
> > Brian
> >
> > To Unsubscribe: send mail to majordomo@FreeBSD.org
> > with "unsubscribe freebsd-smp" in the body of the message
> >
>
--
work: danderse@cs.utah.edu me: angio@pobox.com
University of Utah http://www.angio.net/
Department of Computer Science
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-smp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199811081835.LAA16723>