xfs
[Top] [All Lists]

Re: XFS feature request

To: Austin Gonyou <austin@xxxxxxxxxxxxxxx>
Subject: Re: XFS feature request
From: Michael Loftis <mloftis@xxxxxxxxx>
Date: Wed, 11 Jun 2003 17:29:33 -0600
Cc: XFS List <linux-xfs@xxxxxxxxxxx>
In-reply-to: <1055367128.2003.57.camel@portageek.digitalroadkill.net>
References: <86984947.1054766968@[10.1.2.77]> <1055367128.2003.57.camel@portageek.digitalroadkill.net>
Sender: linux-xfs-bounce@xxxxxxxxxxx
True...There may be situations where having the filesystem degrade into nothing more than a semi-useful source of random entropy for PRNG may be ok :)

--On Wednesday, June 11, 2003 4:32 PM -0500 Austin Gonyou <austin@xxxxxxxxxxxxxxx> wrote:

Given the path of comments here. I'd say that's why the ext2 FS has 3
options for error handling! :)

On Wed, 2003-06-04 at 23:49, Michael Loftis wrote:
Mostly I'd much prefer the option of just going R/O on a filesystem,
living
with the possibility that what I'm *reading* isn't right than a whole
system just going black and having to get onself physically present to
ascertain the problem.

I don't believe that unmounting the filesystem is *always* the right
thing
to do.  It is in some cases, but there are cases where you'd much
rather
the system just went read-only and refused to flush anything off to
it.

--On Wednesday, June 04, 2003 2:02 PM -0600 Andrew Mathews
<andrew_mathews@xxxxxxxxxxxxxxx> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Michael Loftis wrote:
>| This has also recently been a problem for us.  The whole 'something
>| looks a little odd so I'm going to just tank and take the whole
system
>| with me' knee-jerk reaction XFS has is very problematic.  It's the
>| biggest reason I'm moving operations away from it.
>|
>| --On Wednesday, June 04, 2003 20:26 +0200 Seth Mos
<knuffie@xxxxxxxxx>
>| wrote:
>|
>|> Hello,
>|>
>|> Upon a filesystem error the XFS filesystem normally unmounts the
fs and
>|> often takes the box along with it.
>|> ...
>|
>| --
>| Michael Loftis
>| Modwest Sr. Systems Administrator
>| Powerful, Affordable Web Hosting
>|
>
> Not to diminish Set's original request, but to address this
comment...
>
> I guess your term "knee-jerk reaction" is subjective. I'd much
rather
> have a recoverable system than an unrecoverable one, which is
exactly
> what you'll have if the filesystem is put into an unstable state
> (corruption) and isn't intelligent enough to protect itself from
further
> damage. It's also one of the reasons we won't use anything other
than
> XFS in a production environment.
>
> - --
> Andrew Mathews
> -
---------------------------------------------------------------------
> ~  1:56pm  up 5 days,  4:49, 12 users,  load average: 1.16, 1.13,
1.09
> -
---------------------------------------------------------------------
> A statesman is a politician who's been dead 10 or 15 years.
>               -- Harry S. Truman
> - --
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.0.7 (GNU/Linux)
> Comment: Using GnuPG with Netscape - http://enigmail.mozdev.org
>
> iD8DBQE+3lBHidHQ0m/kEssRAhouAJwJPdLuIKyUzmxSAj2nySsT2zCGfgCeO4lf
> 33cLIwDjO7/T/iY8LSDp5qQ=
> =T97e
> -----END PGP SIGNATURE-----
>
>


--
Austin Gonyou <austin@xxxxxxxxxxxxxxx>
Coremetrics, Inc.





<Prev in Thread] Current Thread [Next in Thread>