Subject: Re: raidframe parity: daily and before securelevel
To: Alan Barrett <apb@cequrux.com>
From: Daniel Carosone <dan@geek.com.au>
List: tech-userlevel
Date: 05/18/2007 16:56:47
--AA9g+nFNFPYNJKiL
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Fri, May 18, 2007 at 08:40:26AM +0200, Alan Barrett wrote:
> On Thu, 17 May 2007, Greg Oster wrote:
> > The only reason a parity check should fail is if one of the=20
> > components has errors
>=20
> Oh, did somebody fix the issues with shutdown doing stuff in the wrong
> order, leaving raidframe parity dirty in some cases even without any
> hardware failures?  (ISTR that using swap on raid was one of the easier
> ways of triggering this bug.)

IIUC, as long as you run the shutdown hooks, including swapoff, the
ordering is catered for.  swapoff may not be run by default, because
it can force lots of pageins for memory that's about to be thrown away
anyway, slowing down the reboot for all machines, regardless of raid.

Your point remains valid anyway: an ungraceful shutdown of any kind is
another way to get a bad parity state, even with no component failure.

--
Dan.

--AA9g+nFNFPYNJKiL
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (NetBSD)

iD8DBQFGTU4vEAVxvV4N66cRApFBAJ4ugn3XqhN5zOvMI1niBFA5u9ocJwCgriZE
kJ4efNpr1hyN540SdPYWmTw=
=HYYd
-----END PGP SIGNATURE-----

--AA9g+nFNFPYNJKiL--