---------- Forwarded Message ----------
Date: Wednesday, October 03, 2001 03:38:55 PM +0200
From: Clemens Schwaighofer <[email protected]>
To: Dave Jones <[email protected]>
Subject: Re: [POT] Which journalised filesystem ?
Hello Dave Jones
--On Wednesday, October 03, 2001 02:54:17 PM +0200 you wrote:
> Alan mentioned this was something to do with the IBM hard disk
> having strange write-cache properties that confuse ext3.
> I'm not sure if this has been fixed or not yet, but its enough
> to make me think twice about trying it on the vaio for a while.
I used it succesfully on a VAIO notebook (2.4.10 + ext3) and I had no
troubles. But I had that notebook only for a week, so I couldn't do any
extensible tests, the only thing I did, was cutting the power and seeing if
it recovers. and it did. fast and without problems.
but to the point of that thread. we had reiser FS on a production server
(Fileserver for NFS, Samba & Appletalk) and we nothing but troubles. It was
an 2.2.16 kernel and i dunno witch reiserfs we used. But from this point
forward I dun think I will use it again soon on a production server.
--
"Freiheit ist immer auch die Freiheit des Andersdenkenden"
Rosa Luxemburg, 1871 - 1919
mfg, Clemens Schwaighofer PIXELWINGS Medien AG
Kandlgasse 15/5, A-1070 Wien T: [+43 1] 524 58 50
JETZT NEU! MIT FEWA GEWASCHEN --> http://www.pixelwings.com
---------- End Forwarded Message ----------
--
"Freiheit ist immer auch die Freiheit des Andersdenkenden"
Rosa Luxemburg, 1871 - 1919
mfg, Clemens Schwaighofer PIXELWINGS Medien AG
Kandlgasse 15/5, A-1070 Wien T: [+43 1] 524 58 50
JETZT NEU! MIT FEWA GEWASCHEN --> http://www.pixelwings.com
In article <[email protected]> you wrote:
> but to the point of that thread. we had reiser FS on a production server
> (Fileserver for NFS, Samba & Appletalk) and we nothing but troubles. It was
> an 2.2.16 kernel and i dunno witch reiserfs we used. But from this point
> forward I dun think I will use it again soon on a production server.
Do you had NFS Problems or do you had filesystem problems?
Because NFS interaction with Journaled Filesystems is/was an issue with
those recent kernels, as far as i understand.
Greetings
Bernd
Hello Bernd Eckenfels
--On Wednesday, October 03, 2001 08:01:03 PM +0200 you wrote:
> In article <[email protected]> you wrote:
>> but to the point of that thread. we had reiser FS on a production server
>> (Fileserver for NFS, Samba & Appletalk) and we nothing but troubles. It
>> was an 2.2.16 kernel and i dunno witch reiserfs we used. But from this
>> point forward I dun think I will use it again soon on a production
>> server.
>
> Do you had NFS Problems or do you had filesystem problems?
Filesystem Problems. Massive problems. It went so far, that the system was
so unstable, that I had to reboot it almost everyday.
> Because NFS interaction with Journaled Filesystems is/was an issue with
> those recent kernels, as far as i understand.
I might have came from NFS, AppleTalk, Samba, who knows. But I couldn't go
into detail testing, and I needed to fix it up. I might try ext3 one, cause
I work with it at home and I am quite happy with it, but it's just a home
system not a production enviroment ...
--
"Freiheit ist immer auch die Freiheit des Andersdenkenden"
Rosa Luxemburg, 1871 - 1919
mfg, Clemens Schwaighofer PIXELWINGS Medien AG
Kandlgasse 15/5, A-1070 Wien T: [+43 1] 524 58 50
JETZT NEU! MIT FEWA GEWASCHEN --> http://www.pixelwings.com
Hi,
On Wed, Oct 03, 2001 at 08:01:03PM +0200, Bernd Eckenfels wrote:
>
> Do you had NFS Problems or do you had filesystem problems?
>
> Because NFS interaction with Journaled Filesystems is/was an issue with
> those recent kernels, as far as i understand.
Should be fine with ext3 and XFS. It's not a journaling problem as
much as NFS assuming a particular property of the filesystem.
Resierfs had a particular difficulty with NFS, mainly because the NFS
spec assumes that every file can be looked up by a 64-bit cookie which
doesn't change over reboots, and that's a hard invariant to deal with
when you've only got 32-bit inode numbers in the kernel and when your
filesystem is tree-structured so that the file metadata on disk can
move about. The VFS has been extended a bit in more recent kernels to
allow Reiserfs to give NFS the hints it needs to get the file handles
right.
Cheers,
Stephen
"Stephen C. Tweedie" wrote:
>
> Hi,
>
> On Wed, Oct 03, 2001 at 08:01:03PM +0200, Bernd Eckenfels wrote:
> >
> > Do you had NFS Problems or do you had filesystem problems?
> >
> > Because NFS interaction with Journaled Filesystems is/was an issue with
> > those recent kernels, as far as i understand.
>
> Should be fine with ext3 and XFS. It's not a journaling problem as
> much as NFS assuming a particular property of the filesystem.
>
> Resierfs had a particular difficulty with NFS, mainly because the NFS
> spec assumes that every file can be looked up by a 64-bit cookie which
> doesn't change over reboots, and that's a hard invariant to deal with
> when you've only got 32-bit inode numbers in the kernel and when your
> filesystem is tree-structured so that the file metadata on disk can
> move about. The VFS has been extended a bit in more recent kernels to
> allow Reiserfs to give NFS the hints it needs to get the file handles
> right.
>
> Cheers,
> Stephen
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
NFS and Reiserfs interaction is stable from all reports.
Hans
"Stephen C. Tweedie" wrote:
>
> Hi,
>
> On Wed, Oct 03, 2001 at 08:01:03PM +0200, Bernd Eckenfels wrote:
> >
> > Do you had NFS Problems or do you had filesystem problems?
> >
> > Because NFS interaction with Journaled Filesystems is/was an issue with
> > those recent kernels, as far as i understand.
>
> Should be fine with ext3 and XFS. It's not a journaling problem as
> much as NFS assuming a particular property of the filesystem.
>
> Resierfs had a particular difficulty with NFS, mainly because the NFS
> spec assumes that every file can be looked up by a 64-bit cookie which
> doesn't change over reboots, and that's a hard invariant to deal with
> when you've only got 32-bit inode numbers in the kernel and when your
> filesystem is tree-structured so that the file metadata on disk can
> move about. The VFS has been extended a bit in more recent kernels to
> allow Reiserfs to give NFS the hints it needs to get the file handles
> right.
>
> Cheers,
> Stephen
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
Err, I meant it is stable from all reports for recent kernels:-/.... excuse me.
Hans
Clemens Schwaighofer wrote:
>
> Hello Bernd Eckenfels
>
> --On Wednesday, October 03, 2001 08:01:03 PM +0200 you wrote:
>
> > In article <[email protected]> you wrote:
> >> but to the point of that thread. we had reiser FS on a production server
> >> (Fileserver for NFS, Samba & Appletalk) and we nothing but troubles. It
> >> was an 2.2.16 kernel and i dunno witch reiserfs we used. But from this
> >> point forward I dun think I will use it again soon on a production
> >> server.
> >
> > Do you had NFS Problems or do you had filesystem problems?
>
> Filesystem Problems. Massive problems. It went so far, that the system was
> so unstable, that I had to reboot it almost everyday.
>
> > Because NFS interaction with Journaled Filesystems is/was an issue with
> > those recent kernels, as far as i understand.
>
> I might have came from NFS, AppleTalk, Samba, who knows. But I couldn't go
> into detail testing, and I needed to fix it up. I might try ext3 one, cause
> I work with it at home and I am quite happy with it, but it's just a home
> system not a production enviroment ...
>
> --
> "Freiheit ist immer auch die Freiheit des Andersdenkenden"
> Rosa Luxemburg, 1871 - 1919
> mfg, Clemens Schwaighofer PIXELWINGS Medien AG
> Kandlgasse 15/5, A-1070 Wien T: [+43 1] 524 58 50
> JETZT NEU! MIT FEWA GEWASCHEN --> http://www.pixelwings.com
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
You would find it much more stable today, and I expect that it was NFS plus
reiserfs interaction that was the bug back then for you.
Hans
Hi,
On Fri, Oct 05, 2001 at 02:42:10PM +0400, Hans Reiser wrote:
> > Should be fine with ext3 and XFS. It's not a journaling problem as
> > much as NFS assuming a particular property of the filesystem.
> Err, I meant it is stable from all reports for recent kernels:-/.... excuse me.
Yes, and it's also worth noting that this same NFS assumption will
break exports of _all_ filesystems which don't have simple static
inum/filehandle capabilities. Reiserfs should work just fine now but
the same NFS problem is still present if you do other things such as
trying to re-export a SMB mount as NFS.
Cheers,
Stephen
On Fri, Oct 05, 2001 at 12:25:36PM +0100, Stephen C. Tweedie wrote:
> Hi,
>
> On Fri, Oct 05, 2001 at 02:42:10PM +0400, Hans Reiser wrote:
>
> > > Should be fine with ext3 and XFS. It's not a journaling problem as
> > > much as NFS assuming a particular property of the filesystem.
>
> > Err, I meant it is stable from all reports for recent kernels:-/.... excuse me.
>
> Yes, and it's also worth noting that this same NFS assumption will
> break exports of _all_ filesystems which don't have simple static
> inum/filehandle capabilities. Reiserfs should work just fine now but
> the same NFS problem is still present if you do other things such as
> trying to re-export a SMB mount as NFS.
>
Yes, this issue has been reported many times on the samba mailing lists...
Are there any network filesystems on linux that have the "static
inum/filehandle capabilities"?