Subject: Re: Log area on-disk for the journal
To: Michael van Elst <mlelstv@serpens.de>
From: Bill Studenmund <wrstuden@netbsd.org>
List: tech-kern
Date: 10/23/2006 13:21:33
--p4qYPpj5QlsIQJ0K
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
On Mon, Oct 23, 2006 at 08:59:08PM +0200, Michael van Elst wrote:
> On Mon, Oct 23, 2006 at 11:25:08AM -0700, Jason Thorpe wrote:
> >=20
> > >True, the journal would be lost.
> >=20
> > It's not just the journal -- if you can't recover the journal, you =20
> > can't even be sure that the file system is fsck'able.
>=20
> So how does the rest of the world do it?
They probably have hooks and steps to cover the problems we're talking=20
about.
For what you mentioned Solaris does, we could make journaling a mount=20
option. You create a file system with a journal but leave it empty. Then=20
just do all your operations meta-data-synchronous, and all's well. If it=20
was journaled before but isn't now, just roll the journal forward, and=20
you're fine.
Take care,
Bill
--p4qYPpj5QlsIQJ0K
Content-Type: application/pgp-signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (NetBSD)
iD8DBQFFPSRNWz+3JHUci9cRAjxfAJ9Izi2V8Ul6b/l6Wq8kn8SLh9UlmgCfVZld
Xp/hONLSuw2biuTWQ79TAmw=
=gFYY
-----END PGP SIGNATURE-----
--p4qYPpj5QlsIQJ0K--