Re: [Woody] Mounting LVM-snapshots with ext3 + ACL + user_xattr-Patch
Paul Morgan <[EMAIL PROTECTED]> wrote: > This is a well known issue. lvm snapshots don't work in the 2.4 > kernel series. The reason is that ext3 needs to lock and flush the > ext3 journal *before* taking the snapshot. Does that imply, that lvm snapshots should work with 2.6.0 kernels? When I try the following (ok its xfs but ...) xfs_freeze -f /home lvcreate --size 250m -s -n snap1 /dev/vg00/home_lv xfs_freeze -u /home I get this error: Rounding up size to full physical extend 252.00 MB device-mapper: error adding target to table device-mapper ioctl cmd 9 failed: Invalid argument Couldn't load device 'vg00-snap1'. Problem reactivating origin home_lv ?! With a patched Debian kernel that worked ok with 2.4.22 (IIRC). by Töns -- There is no safe distance. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: [Woody] Mounting LVM-snapshots with ext3 + ACL + user_xattr-Patch
On Mon, 05 Jan 2004 15:19:43 +0100, Oliver Schade wrote: > Dear all, > > > I am using a Woody system, running a Vanilla 2.4.23 kernel from > ftp.kernel.org and the ACL+XATTR-Patch vom acl.bestbits.at. > > Working with ext3 and ACLs on a logical volume works fine, fast and > stable. But I have some troubles mounting a snapshot of a live lvm- > volume named /dev/RAID5/TEMP: > > fsb01:~# mount | grep TEMP > /dev/RAID5/TEMP on /mnt/berlin/TEMP type ext3 (rw,acl,user_xattr) > > To create a snapshot I used: > > fsb01:~# lvcreate --size 250m --snapshot --name TEMP1500 /dev/RAID5/TEMP > [... snipped ...] > lvcreate -- logical volume "/dev/RAID5/TEMP1500" successfully created > > The special device has been successfully created: > > fsb01:~# ls -la /dev/RAID5/TEMP1500 > brw-rw1 root disk 58, 9 Jan 5 15:20 /dev/RAID5/TEMP1500 > > However, mounting this snapshot fails: > > fsb01:~# mount -o ro /dev/RAID5/TEMP1500 /mnt/berlin/Snapshots/TEMP > mount: wrong fs type, bad option, bad superblock on /dev/RAID5/TEMP1500, >or too many mounted file systems > This is a well known issue. lvm snapshots don't work in the 2.4 kernel series. The reason is that ext3 needs to lock and flush the ext3 journal *before* taking the snapshot. There is a VFS locking patch for earlier kernels; I have modified it for my kernel 2.4.22. I don't see it on the debian distro, but on googling I see a VFS Locking patch included in LVM 1.0.8 here: http://fresh.t-systems-sfr.com/cgi-bin/warex?linux/src/lvm_1.0.8.tar.gz:a/LVM/1.0.8/PATCHES/linux-2.4.20-VFS-lock.patch At first glance, it looks the same as the patch which I have modified for lvm 1.0.7 and kernel 2.4.22, which is what I would expect. The way I modded the patch was to look in the various sources for the code to be patched and change the line numbers in the patch to be consistent. "man patch" for how to apply it. Seems to work OK, but I'm not recommending it to anyone but myself, so, if it breaks, you get to keep all the pieces. -- paul Programming without a hex editor is like watchmaking without a hammer. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: [Woody] Mounting LVM-snapshots with ext3 + ACL + user_xattr-Patch
Oliver Schade <[EMAIL PROTECTED]> wrote: > However, mounting this snapshot fails: > > fsb01:~# mount -o ro /dev/RAID5/TEMP1500 /mnt/berlin/Snapshots/TEMP > mount: wrong fs type, bad option, bad superblock on /dev/RAID5/TEMP1500, >or too many mounted file systems [...] > Any help for snapshoting and mounting of a live LVM-systems together > with ACLs are highly welcome. For XFS the LVM-Howto suggests: mount -o nouuid,ro /dev/ops/dbbackup /mnt/ops/dbbackup Have you tried that - just in case? by Töns -- There is no safe distance. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
[Woody] Mounting LVM-snapshots with ext3 + ACL + user_xattr-Patch
Dear all, I am using a Woody system, running a Vanilla 2.4.23 kernel from ftp.kernel.org and the ACL+XATTR-Patch vom acl.bestbits.at. Working with ext3 and ACLs on a logical volume works fine, fast and stable. But I have some troubles mounting a snapshot of a live lvm- volume named /dev/RAID5/TEMP: fsb01:~# mount | grep TEMP /dev/RAID5/TEMP on /mnt/berlin/TEMP type ext3 (rw,acl,user_xattr) To create a snapshot I used: fsb01:~# lvcreate --size 250m --snapshot --name TEMP1500 /dev/RAID5/TEMP [... snipped ...] lvcreate -- logical volume "/dev/RAID5/TEMP1500" successfully created The special device has been successfully created: fsb01:~# ls -la /dev/RAID5/TEMP1500 brw-rw1 root disk 58, 9 Jan 5 15:20 /dev/RAID5/TEMP1500 However, mounting this snapshot fails: fsb01:~# mount -o ro /dev/RAID5/TEMP1500 /mnt/berlin/Snapshots/TEMP mount: wrong fs type, bad option, bad superblock on /dev/RAID5/TEMP1500, or too many mounted file systems Followed the discussion on RedHat Bugzilla (https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=88115) I tried to specify a superblock using the mount-options -o ro,sb=32768 and -o ro,sb=819200 but failed also. Both superblock copies are identical to /dev/RAID5/TEMP. An additional "-t ext3" failed also. Any help for snapshoting and mounting of a live LVM-systems together with ACLs are highly welcome. Thanks, Oliver -- pro|business Berlin AG [EMAIL PROTECTED] Potsdamer Platz 11 http://www.probusiness.de/ 10785 BerlinTel: +49 030 259 378-0 Germany Fax: +49 030 259 378-22 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]