Subject: Re: Why not softdep per default?
To: Karsten Kruse <firstname.lastname@example.org>
From: Bill Studenmund <email@example.com>
Date: 04/05/2005 16:14:19
Content-Type: text/plain; charset=us-ascii
On Sun, Apr 03, 2005 at 07:02:02PM +0200, Karsten Kruse wrote:
> J Chapman Flack wrote:
> >>:( =3D may be broken, worst thing: even a manual fsck can't repair it
> >>:) =3D should be ok, worst thing: some inclaimed inodes/blocks
> >> softdep, writebackcache on | no softdep, writebackcache on
> >>powerfailure :( | :(
> >>reset :) | :(
> >>kernel panic :) | :(
> >>The filesystem with softdep wins if writeback cache is on (wich
> >>is default).
Through out this whole thread, you have asserted that life is safer with=20
softdeps on than with it off. I do not believe that is true. I believe=20
that life is _faster_ with softdeps on, not safer.
File system consistency is maintained (in ffs on NetBSD) by ensuring that
certain writes happen before others. softdeps does this with async
ordering. By default, ffs does this with synchronous writes. The issue,
though, is that they both follow the same ordering.
Since they have the same ordering, I don't see how one is safer than the=20
Note: I have compared life to normal ffs. If you were instead to compare=20
softdeps with ffs async, then you would be correct about the safety=20
increase. But async is no more the default option on NetBSD than is=20
> > Now, if you were prepared to present experimental data somehow showing
> > that softdep preserved integrity better on writeback caches in practice,
> > it would be interesting to look at, and speculate why (perhaps it is
> > possible that the overall reduction in traffic from softdep could have
> > such a serendipitous effect - or perhaps not). When you cited the
> > 'experiments' paper I thought that's what it might be, until I looked
> > and saw it did not pertain to ffs or softdep.
> Ok, so i have to make some tests. Here is the plan:
As above, I do not think such tests are needed.=20
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (NetBSD)
-----END PGP SIGNATURE-----