Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 6 Apr 2012 15:00:06 +0200
From:      Pawel Jakub Dawidek <pjd@FreeBSD.org>
To:        "Desai, Kashyap" <Kashyap.Desai@lsi.com>
Cc:        "freebsd-fs@freebsd.org" <freebsd-fs@freebsd.org>, "freebsd-scsi@freebsd.org" <freebsd-scsi@freebsd.org>, "Kenneth D.Merry" <ken@FreeBSD.org>, "McConnell, Stephen" <Stephen.McConnell@lsi.com>
Subject:   Re: Kernel crash at "softdep_deallocate_dependencies"
Message-ID:  <20120406130006.GC1336@garage.freebsd.pl>
In-Reply-To: <B2FD678A64EAAD45B089B123FDFC3ED72B96EF90FE@inbmail01.lsi.com>
References:  <B2FD678A64EAAD45B089B123FDFC3ED72B96EF90FE@inbmail01.lsi.com>

next in thread | previous in thread | raw e-mail | index | archive | help

--KN5l+BnMqAQyZLvT
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Fri, Apr 06, 2012 at 03:31:39PM +0530, Desai, Kashyap wrote:
> Hi, Thanks in advance for looking this query and hoping for some help fro=
m File system experts.
>=20
> We have a RAID 0 volume which has partition (created using sysinstall fdi=
sk option).
> While IOs are in progress removing one of the volume member kernel panic =
is hit with the following messages
>=20
> g_vfs_done():(da0:da0s1d[WRITE(offset=3D6358872064, length=3D2048)]mpslsi=
0:0:error =3D 6
> 0:/home: got error 6 while accessing filesystem
> 0): panic: softdep_deallocate_dependencies: unrecovered I/O error
> lost device
>=20
> Note:
> 1. The issue is also seen on a RAID 0 volume which does not have a partit=
ion on it.
> 2. Issue was observed on both SAS and SATA drives.
> 3. When we send IOs to the driver without FS (using "dd" command), kernel=
 panic never seen.
>=20
>=20
> I have searched on this topic and looks like something wrong with FS. _bu=
t_ I don't have any trigger to support that this is not Driver issue.

This might be lame error handling on the FS side, but FS is not here to
blame. You get I/O error from device below. In case of RAID0 you have no
redundancy, so you cannot expect anything good by removing one of its
components.

Using "dd" doesn't trigger kernel panic, because I/O error is handled by
userland process (it exits).

> When we tried below options: (disable Journal on FS) Things does not chan=
ge. We still see kernel panic.
>=20
> umount <filesystem>
> tunefs -j disable <filesystem>
> mount <filesystem>
> cd <filesystem>
> rm .sujournal
>=20
> Any thoughts ?

What behaviour would you expect when your RAID0 volume dies? The best
thing to do here would be to either stop all I/Os until the component is
back or forcibly unmount the file system, but both options are probably
hard to get right.

--=20
Pawel Jakub Dawidek                       http://www.wheelsystems.com
FreeBSD committer                         http://www.FreeBSD.org
Am I Evil? Yes, I Am!                     http://tupytaj.pl

--KN5l+BnMqAQyZLvT
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.18 (FreeBSD)

iEYEARECAAYFAk9+6NYACgkQForvXbEpPzTujQCg3rzzLyyVrVc/UtdrmFbb9PHA
nxsAoOo1fZRtJoYUw+lIM2nKWqT9Xm0p
=E9dp
-----END PGP SIGNATURE-----

--KN5l+BnMqAQyZLvT--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20120406130006.GC1336>