Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 11 Dec 2001 20:11:26 +0000
From:      ian j hart <ianjhart@ntlworld.com>
To:        "stable@FreeBSD.ORG" <stable@freebsd.org>
Subject:   Re: 4.4-STABLE crashes - suspects new ata-driver over wd-drivers
Message-ID:  <3C16686D.7FDC2D76@ntlworld.com>
References:  <20011207133314.A20037@3docs.com> <20011207214323.F19268@freeway.dcfinc.com> <3C13B4C8.EB277D18@ntlworld.com> <3C13E1C4.661CB61A@ntlworld.com>

next in thread | previous in thread | raw e-mail | index | archive | help
ian j hart wrote:
> 
> ian j hart wrote:
> >
> > "Chad R. Larson" wrote:
> > >
> > > On Fri, Dec 07, 2001 at 01:33:15PM -0800, Brady Montz wrote:
> > > > Yeah, I'm using soft updates too.  My crashes are generally the
> > > > same as Richards - no panic, just a freeze.  Except my screen
> > > > doesn't go blank.
> > >
> > > For what it's worth, I'm using soft updates on a web server that gets
> > > steady if not heavy use.  Built from RELENG_4_3, and no problems at
> > > all.
> > >
> > >         -crl
> > > --
> > > Chad R. Larson (CRL15)   602-953-1392   Brother, can you paradigm?
> > > chad@dcfinc.com         chad@larsons.org          larson1@home.com
> > > DCF, Inc. - 14623 North 49th Place, Scottsdale, Arizona 85254-2207
> > >
> > > To Unsubscribe: send mail to majordomo@FreeBSD.org
> > > with "unsubscribe freebsd-stable" in the body of the message
> >
> > Me 2 :(
> >
> > I have a total lockup, screen is not blank (Matrox G400).
> >
> > I turned off soft updates and did a boot -v and got some
> > console messages, so this is worth a try. Unfortunately the
> > messages don't make it to the logs, presumably because the
> > disk and/or disk subsystem is fubar'd. The one time a got
> > a spontaineous reboot I was out of the room making coffee
> > (typical).
> >
> > Anyway the messages are something like
> > ad0: READ command timeout tag=0 serv=0 - resetting
> > ata0: resetting devices .. done
> >
> > It's not always the same drive.
> >
> > There were also some of my favorite "UDMA ICRC" errors, but
> > I didn't catch those. For those with long memories this is
> > the same box I've had UDMA problems with before (numerous
> > posts with UDMA ICRC in subject) but it's been well behaved
> > since early July. Maybe I haven't pushed it hard enough.
> > I also got one instance of "unexpected soft update inconsisency"
> > while fscking. Maybe this is to be expected if the drive "just
> > dies".
> >
> > What's interesting is the behavior seems to have changed. On
> > previous occasions the driver would keep resetting and then
> > drop to pio mode. Now it seems to lock after the first reset.
> > I'll try to confirm this behavior.
> >
> > I set pio mode on all drives and I managed to complete my
> > torture test.
> >
> > One more thing. Sometimes there's a clunk from the drive{s)
> > when it dies. Parking the heads?
> >
> > FWIW -
> > VIA ATA33 controller
> > 4x UDMA 66 drives
> > vinum mirror /var
> > vinum mirrored stripes /usr
> >
> 
> Drat, spoke too soon.
> 
> soft updates on, dma off. Hang (in kde) followed by black
> screen and reboot. This time vinum died on startup and I
> had an anxious 10 minutes starting all the subdisks.
> 
> I'd better test the memory. Then I'll try booting from
> the backup root in case ad0 is toast. I guess duff hardware is
> looking more likely.
> 
> I noticed some UDMA errors when rebooting from single
> user, which failed to sync 1 block. Of course these scroll
> off screen too quick to be readable, but the "head parking"
> noise was again apparent. APM is disabled.
> 
> --
> ian j hart
> 
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-stable" in the body of the message

Update:

I couldn't prove the memory faulty but I did discover a useful
factlet which was missed out of the M/B handbook. Apparently
the VIA MVP3 chipset only supports 16MxN RAM when you have
more than 64Mb. This is not what I had. I updated the BIOS and
sure enough the board failed to detect all the RAM. I've
swapped it out. Maybe some update finally tickled the "feature"
hard enough to cause a panic.

I also did a full build, and this seems to have fixed some
weirdness with md0. Either I cvsup'd at a bad time or (more
likely) I fluffed the mergemester.

I'll thrash the bejesus out of the drives and see what happens.

-- 
ian j hart

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3C16686D.7FDC2D76>