From: Klaus Strebel <klaus.strebel@gmx.net>
To: Dave N <mutex1@yahoo.com>
Cc: xfs@oss.sgi.com
Subject: Re: What's wrong with XFS?
Date: Mon, 08 Jan 2007 16:24:23 +0100 [thread overview]
Message-ID: <45A26227.2080907@gmx.net> (raw)
In-Reply-To: <20070108144549.GA12073@tuatara.stupidest.org>
Chris Wedgwood schrieb:
> On Mon, Jan 08, 2007 at 05:13:12AM -0800, Dave N wrote:
>
>> KDE, I was surprised to find out that all my desktop icons were
>> messed up all over the place.
>
> KDE made assumptions which are not only not true on linux but not true
> elsewhere either. Last I checked KDE dealt with the common cases that
> were problematic much better now.
>
>> The other time, again power outage, only this time I was working on
>> a small text file. Booted up again only to find out that the file I
>> was working on contained garbage and I had to start all over again.
>
> The file should not have contained garbage. Also, if you
> open+truncate+write a file it should be flushed very soon after close
> these days, the window is fairly small now.
>
>> I also heard that XFS depends heavily on the application side for
>> its data-integrity. XFS "thinks" that the application will use the
>> proper calls when writing to disk. What???? How is it the task of
>> the application to ensure the safety of your files???
>
> It's always been that way, for many many years, even before Linux
> existed. If you want your applictions to be portable and reliable
> then you have to do do it right.
>
> MTAs are a good example of applications which typically get this right
> because people case about lost email and the authors typically take
> some effort into make sure it's right.
>
>> IMO, programs are there to provide the tools to be productive, NOT
>> to ensure the data safety of your files, that's the task of the file
>> system. Even MySQL provides me with better data-integrity here.
>
> Does MySQL allow me to read or write 100s of MB/s continuously on
> cheap hardware (for not so cheap hardware I could ask 7GB/s).
>
>> Why oh why such a beautiful file system like XFS is so terrible at
>> data-integrity?
>
> There is a cost to full data journalling. Personally even with ext3 I
> find the impact of this high enough I don't use it.
>
>> Look what Sun Microsystems did with their new ZFS file
>> system... full atomicity, CRC checksumming and other features to
>> ensure data-integrity...
>
> You could argue XFS is showing it's age, it's far from a new
> filesystem these days.
>
> ZFS is a very different animal to most traditional filesystems.
>
>> why can't XFS have such things?
>
> Because the realities of life sometime collide with what people want
> ideally.
>
> Linux can't have ZFS for licensing reasons but you can have Solaris
> with ZFS: http://opensolaris.org/os/downloads/on/
>
>
FYI,
just found this <http://developer.berlios.de/projects/zfs-fuse/>
;-)
Klaus
--
Mit freundlichen Grüssen / best regards
Klaus Strebel, Dipl.-Inform. (FH), mailto:klaus.strebel@gmx.net
/"\
\ / ASCII RIBBON CAMPAIGN
X AGAINST HTML MAIL
/ \
next prev parent reply other threads:[~2007-01-08 15:25 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-01-08 13:13 What's wrong with XFS? Dave N
2007-01-08 14:41 ` Klaus Strebel
2007-01-08 14:45 ` Chris Wedgwood
2007-01-08 15:24 ` Klaus Strebel [this message]
2007-01-08 14:49 ` Olaf Fraczyk
2007-01-08 15:35 ` Joe Bacom
2007-01-08 14:50 ` Olaf Frączyk
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=45A26227.2080907@gmx.net \
--to=klaus.strebel@gmx.net \
--cc=mutex1@yahoo.com \
--cc=xfs@oss.sgi.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox