[linux-lvm] [catch22] lvm vs reiserfs

[linux-lvm] [catch22] lvm vs reiserfs

[Ed Tomlinson]

Hi,

I decided to really try out some of the features built into 2.4.  Using a 
2.4.2-ac20 kernel I my root on /dev/lv/root as a reiserfs filesystem.  So
far so good.  Now I want to back it up.  So I created a snapshot with
lvcreate -L 50 -s -n snap /dev/lv/root.  This works too.  Problem happens
when I want to mount /dev/lv/snap to back it up logically.  At this point
reiserfs wants to replay its journal....  Of course it cannot (snapshot=ro).

How can this be resolved?  Is there a way to sync root just after the lvm
data gets updated in /etc but before the snapshot starts?  How else might
this be fixed (is it fixed already somewhere)?

TIA
Ed Tomlinson

Re: [linux-lvm] [catch22] lvm vs reiserfs

[Chris Mason]

On Thursday, March 22, 2001 02:46:31 PM -0500 Ed Tomlinson
<tomlins@cam.org> wrote:

> Hi,
> 
> I decided to really try out some of the features built into 2.4.  Using a 
> 2.4.2-ac20 kernel I my root on /dev/lv/root as a reiserfs filesystem.  So
> far so good.  Now I want to back it up.  So I created a snapshot with
> lvcreate -L 50 -s -n snap /dev/lv/root.  This works too.  Problem happens
> when I want to mount /dev/lv/snap to back it up logically.  At this point
> reiserfs wants to replay its journal....  Of course it cannot
> (snapshot=ro).
> 
> How can this be resolved?  Is there a way to sync root just after the lvm
> data gets updated in /etc but before the snapshot starts?  How else might
> this be fixed (is it fixed already somewhere)?

Go to ftp.sistina.com, and grab 0.9.1_beta6.  Apply
linux-2.4.2-VFS-lock.patch that is included in PATCHES directory after
untarring.  This provides the rest of the lockfs API I coded with the LVM
guys.

-chris

Re: [linux-lvm] [catch22] lvm vs reiserfs

[Ed Tomlinson]

On Thursday 22 March 2001 15:10, Chris Mason wrote:
> On Thursday, March 22, 2001 02:46:31 PM -0500 Ed Tomlinson
>
> <tomlins@cam.org> wrote:
> > Hi,
> >
> > I decided to really try out some of the features built into 2.4.  Using a
> > 2.4.2-ac20 kernel I my root on /dev/lv/root as a reiserfs filesystem.  So
> > far so good.  Now I want to back it up.  So I created a snapshot with
> > lvcreate -L 50 -s -n snap /dev/lv/root.  This works too.  Problem happens
> > when I want to mount /dev/lv/snap to back it up logically.  At this point
> > reiserfs wants to replay its journal....  Of course it cannot
> > (snapshot=ro).
> >
> > How can this be resolved?  Is there a way to sync root just after the lvm
> > data gets updated in /etc but before the snapshot starts?  How else might
> > this be fixed (is it fixed already somewhere)?
>
> Go to ftp.sistina.com, and grab 0.9.1_beta6.  Apply
> linux-2.4.2-VFS-lock.patch that is included in PATCHES directory after
> untarring.  This provides the rest of the lockfs API I coded with the LVM
> guys.

I updated to ac21 with the above patch.  I am using the lvm utilities from:

deb ftp://ftp.sistina.com/pub/LVM/0.9/binaries/debian unstable main

Now I can mount the snapshot.  Thanks.

One loose end, is the journal-1777 message below something to worry about?  

oscar# lvcreate -L 50 -s -n snap /dev/lv/root
lvcreate -- rounding up size to physical extent boundary "52 MB"
lvcreate -- WARNING: the snapshot must be disabled if it gets full
lvcreate -- INFO: using default snapshot chunk size of 64 KB for "/dev/lv/snap"
journal-1777: buffer 16 bad state !PREPARED !LOCKED !DIRTY !JDIRTY_WAIT
lvcreate -- doing automatic backup of "lv"
lvcreate -- logical volume "/dev/lv/snap" successfully created

TIA,
Ed Tomlinson

[linux-lvm] Re: [catch22] lvm vs reiserfs

[Ed Tomlinson]

On Thursday 22 March 2001 16:06, Ed Tomlinson wrote:

Chris et all,

Looks like reiserfs snapshots are broken with beta6 and ac22.
I had to upgrade to beta6 because of other problems (vgextend 
caused vgscan to fail...).  Now the snapshots are dead again.  
Note the VFS patch is still applied...

oscar# lvcreate -L 1G -s -n snap /dev/lv/root /dev/hda3
lvcreate -- WARNING: the snapshot must be disabled if it gets full
lvcreate -- INFO: using default snapshot chunk size of 64 KB for "/dev/lv/snap"
lvcreate -- doing automatic backup of "lv"
lvcreate -- logical volume "/dev/lv/snap" successfully created
 
oscar# mount /dev/lv/snap /snap -treiserfs -oro
reiserfs: checking transaction log (device 3a:04) ...
clm-2076: device is readonly, unable to replay log
Replay Failure, unable to mount
reiserfs_read_super: unable to initialize journal space
mount: wrong fs type, bad option, bad superblock on /dev/lv/snap,
       or too many mounted file systems

Ideas?
Ed

> On Thursday 22 March 2001 15:10, Chris Mason wrote:
> > On Thursday, March 22, 2001 02:46:31 PM -0500 Ed Tomlinson
> >
> > <tomlins@cam.org> wrote:
> > > Hi,
> > >
> > > I decided to really try out some of the features built into 2.4.  Using
> > > a 2.4.2-ac20 kernel I my root on /dev/lv/root as a reiserfs filesystem.
> > >  So far so good.  Now I want to back it up.  So I created a snapshot
> > > with lvcreate -L 50 -s -n snap /dev/lv/root.  This works too.  Problem
> > > happens when I want to mount /dev/lv/snap to back it up logically.  At
> > > this point reiserfs wants to replay its journal....  Of course it
> > > cannot
> > > (snapshot=ro).
> > >
> > > How can this be resolved?  Is there a way to sync root just after the
> > > lvm data gets updated in /etc but before the snapshot starts?  How else
> > > might this be fixed (is it fixed already somewhere)?
> >
> > Go to ftp.sistina.com, and grab 0.9.1_beta6.  Apply
> > linux-2.4.2-VFS-lock.patch that is included in PATCHES directory after
> > untarring.  This provides the rest of the lockfs API I coded with the LVM
> > guys.
>
> I updated to ac21 with the above patch.  I am using the lvm utilities from:
>
> deb ftp://ftp.sistina.com/pub/LVM/0.9/binaries/debian unstable main
>
> Now I can mount the snapshot.  Thanks.
>
> One loose end, is the journal-1777 message below something to worry about?
>
> oscar# lvcreate -L 50 -s -n snap /dev/lv/root
> lvcreate -- rounding up size to physical extent boundary "52 MB"
> lvcreate -- WARNING: the snapshot must be disabled if it gets full
> lvcreate -- INFO: using default snapshot chunk size of 64 KB for
> "/dev/lv/snap" journal-1777: buffer 16 bad state !PREPARED !LOCKED !DIRTY
> !JDIRTY_WAIT lvcreate -- doing automatic backup of "lv"
> lvcreate -- logical volume "/dev/lv/snap" successfully created
>
> TIA,
> Ed Tomlinson

Re: [linux-lvm] Re: [catch22] lvm vs reiserfs

[Luca Berra]

On Thu, Mar 22, 2001 at 11:33:26PM -0500, Ed Tomlinson wrote:
> oscar# mount /dev/lv/snap /snap -treiserfs -oro
why readonly?

> reiserfs: checking transaction log (device 3a:04) ...
> clm-2076: device is readonly, unable to replay log
> Replay Failure, unable to mount
> reiserfs_read_super: unable to initialize journal space
> mount: wrong fs type, bad option, bad superblock on /dev/lv/snap,
>        or too many mounted file systems

-- 
Luca Berra -- bluca@comedia.it
        Communication Media & Services S.r.l.
 /"\
 \ /     ASCII RIBBON CAMPAIGN
  X        AGAINST HTML MAIL
 / \

Re: [linux-lvm] Re: [catch22] lvm vs reiserfs

[josv]

Snapshots are by their very nature read-only animals. Their internal
organisation is such that they can not easily be written to.
Mounting it readonly informs the file system of this fact.

It seems from the output that the ReiserFS code wants to replay
the journal, which leads me to believe that the file system was
not properly sync'ed before the snapshot was taken. Maybe the
VFS_lock patch needs to be updated?

++Jos

On Fri, Mar 23, 2001 at 09:19:53AM +0100, Luca Berra wrote:
> On Thu, Mar 22, 2001 at 11:33:26PM -0500, Ed Tomlinson wrote:
> > oscar# mount /dev/lv/snap /snap -treiserfs -oro
> why readonly?
> 
> > reiserfs: checking transaction log (device 3a:04) ...
> > clm-2076: device is readonly, unable to replay log
> > Replay Failure, unable to mount
> > reiserfs_read_super: unable to initialize journal space
> > mount: wrong fs type, bad option, bad superblock on /dev/lv/snap,
> >        or too many mounted file systems
> 
> -- 
> Luca Berra -- bluca@comedia.it
>         Communication Media & Services S.r.l.
>  /"\
>  \ /     ASCII RIBBON CAMPAIGN
>   X        AGAINST HTML MAIL
>  / \
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm

Re: [linux-lvm] Re: [catch22] lvm vs reiserfs

[Ed Tomlinson]

On Thursday 22 March 2001 23:33, Ed Tomlinson wrote:

Hi,

Looks like the generated lvm patch stomps all over the
VFS locking stuff...  Guess the correct procedure is
-R VFS, patch lvm, patch VFS.  Will test tonight

Ed

> On Thursday 22 March 2001 16:06, Ed Tomlinson wrote:
>
> Chris et all,
>
> Looks like reiserfs snapshots are broken with beta6 and ac22.
> I had to upgrade to beta6 because of other problems (vgextend
> caused vgscan to fail...).  Now the snapshots are dead again.
> Note the VFS patch is still applied...
>
> oscar# lvcreate -L 1G -s -n snap /dev/lv/root /dev/hda3
> lvcreate -- WARNING: the snapshot must be disabled if it gets full
> lvcreate -- INFO: using default snapshot chunk size of 64 KB for
> "/dev/lv/snap" lvcreate -- doing automatic backup of "lv"
> lvcreate -- logical volume "/dev/lv/snap" successfully created
>
> oscar# mount /dev/lv/snap /snap -treiserfs -oro
> reiserfs: checking transaction log (device 3a:04) ...
> clm-2076: device is readonly, unable to replay log
> Replay Failure, unable to mount
> reiserfs_read_super: unable to initialize journal space
> mount: wrong fs type, bad option, bad superblock on /dev/lv/snap,
>        or too many mounted file systems
>
> Ideas?
> Ed
>
> > On Thursday 22 March 2001 15:10, Chris Mason wrote:
> > > On Thursday, March 22, 2001 02:46:31 PM -0500 Ed Tomlinson
> > >
> > > <tomlins@cam.org> wrote:
> > > > Hi,
> > > >
> > > > I decided to really try out some of the features built into 2.4. 
> > > > Using a 2.4.2-ac20 kernel I my root on /dev/lv/root as a reiserfs
> > > > filesystem. So far so good.  Now I want to back it up.  So I created
> > > > a snapshot with lvcreate -L 50 -s -n snap /dev/lv/root.  This works
> > > > too.  Problem happens when I want to mount /dev/lv/snap to back it up
> > > > logically.  At this point reiserfs wants to replay its journal.... 
> > > > Of course it cannot
> > > > (snapshot=ro).
> > > >
> > > > How can this be resolved?  Is there a way to sync root just after the
> > > > lvm data gets updated in /etc but before the snapshot starts?  How
> > > > else might this be fixed (is it fixed already somewhere)?
> > >
> > > Go to ftp.sistina.com, and grab 0.9.1_beta6.  Apply
> > > linux-2.4.2-VFS-lock.patch that is included in PATCHES directory after
> > > untarring.  This provides the rest of the lockfs API I coded with the
> > > LVM guys.
> >
> > I updated to ac21 with the above patch.  I am using the lvm utilities
> > from:
> >
> > deb ftp://ftp.sistina.com/pub/LVM/0.9/binaries/debian unstable main
> >
> > Now I can mount the snapshot.  Thanks.
> >
> > One loose end, is the journal-1777 message below something to worry
> > about?
> >
> > oscar# lvcreate -L 50 -s -n snap /dev/lv/root
> > lvcreate -- rounding up size to physical extent boundary "52 MB"
> > lvcreate -- WARNING: the snapshot must be disabled if it gets full
> > lvcreate -- INFO: using default snapshot chunk size of 64 KB for
> > "/dev/lv/snap" journal-1777: buffer 16 bad state !PREPARED !LOCKED !DIRTY
> > !JDIRTY_WAIT lvcreate -- doing automatic backup of "lv"
> > lvcreate -- logical volume "/dev/lv/snap" successfully created
> >
> > TIA,
> > Ed Tomlinson
>
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm