Subject: Re: Corruption on disk slices > 4 gigs
To: John Klos <john@sixgirls.org>
From: Brian Chase <vaxzilla@jarai.org>
List: port-vax
Date: 12/11/2002 05:00:07
On Wed, 11 Dec 2002, John Klos wrote:

> I posted something about disk corruption issues last month:
> http://mail-index.netbsd.org/port-vax/2002/11/22/0003.html
>
> Now I've determined that this ONLY happens when I create disk slices which
> are greater than 4 gigs in size. I tried slicing my disk into a 4 gig and
> a 4.5 gig, and the 4.5 gig partition showed the same behaviour.
>
> Now that I have two 4 gig slices and a 1/2 gig slice, all is well (and
> bulk package building will start and not stop because of a filesystem
> related kernel panic!)
>
> Does anyone have a clue about what might have changed? NetBSD 1.5 used
> a 9 gig partition just fine, and I build many, many packages on it,
> and served almost 400 gigs of web traffic a month for a while without
> problems.

I've had very similar problems with SCSI drives of mine that are larger
than 4GB (4096MB).  I've a few slightly larger than 4GB drives where
this had bit me recently.  I only noticed the problem when I temporarily
swapped in a 1.5.3 boot drive after having built some larger than 4GB
filesystems on my other disks.  The 1.5.3 install didn't like those
large 1.6 newfs'ed filesystems at all.  On reboot, fsck marked them as
being dirty and then *completely* erased the first filesystem before
I realized something was seriously wrong. (Sadness).

I swapped out most of those drives for smaller ones; I'm glad to hear
that it might be a filesystem size issue and not a drive size issue.
I do still have one >4GB filesystem, with hardly any data on it at the
moment.  I've bounced the box a few times under 1.6 trying out
various kernel builds without incident, but I'll try copying some
unimportant data over to it to see if I can get it to misbehave.

This is on a MicroVAX 3100 KA41-D.

-brian.