From owner-freebsd-current  Sun Dec  1 07:08:12 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id HAA17297
          for current-outgoing; Sun, 1 Dec 1996 07:08:12 -0800 (PST)
Received: from ravenock.cybercity.dk (disn53.cybercity.dk [194.16.57.53])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id HAA17292;
          Sun, 1 Dec 1996 07:08:03 -0800 (PST)
Received: (from sos@localhost) by ravenock.cybercity.dk (8.8.3/8.7.3) id QAA23315; Sun, 1 Dec 1996 16:10:26 +0100 (MET)
Message-Id: <199612011510.QAA23315@ravenock.cybercity.dk>
Subject: Re: Corrupted CVS updates
To: dyson@freebsd.org (John S. Dyson)
Date: Sun, 1 Dec 1996 16:10:26 +0100 (MET)
From: "Soren Schmidt" <sos@ravenock.cybercity.dk>
Cc: sos@freebsd.org, current@freebsd.org
In-Reply-To: <32A19869.41C67EA6@freebsd.org> from "John S. Dyson" at Dec 1, 96 09:38:33 am
From: sos@freebsd.org
Reply-to: sos@freebsd.org
X-Mailer: ELM [version 2.4 PL25 ME8b]
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
Sender: owner-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

In reply to John S. Dyson who wrote:
> 
> Soren Schmidt wrote:
> 
> > 
> > Hmm, I have seen various problems with some of the bigger CTM
> > patches (esp. those that touch all files), this triggers our
> > old well known (but not understood) vm problem, that manifests
> > itself as an bogus diskblock (one page), that will disappear if
> > the disk is unmounted & then mounted again 95% of the times, the
> > last 5% that invalid block will have been written back to disk :(
> > It seems that my systems has been suffering pretty bad from this
> > over the last ~14 days or so. I have one workaround for this:
> > mount the disks via nfs and do the update from another machine,
> > this fixes the problem, but is dog slow....
> > 
> Sorry for my Netscape reply -- only mailer that I have right
> now that supports mime...

:)

> Could you tell me if you are mounting your filesystems async?
> I am looking at the problem (in filesystem mode -- unfortunately
> meant to be in documentation mode.)

Well, its almost beyond use when I do NOT mount them async, async
makes the probelm a lot less worse...
Also this only happens on the really large patches, so size is
definatly a parameter in this...

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Søren Schmidt               (sos@FreeBSD.org)               FreeBSD Core Team
                Even more code to hack -- will it ever end
..