As I was shrinking the filesystem during my last post, I wondered if there is an easier way to undo LVM changes. I then ran into vgcfgrestore/vgcfgbackup and I decided to give it a try.

Setup a Test LVM Configuration

So I had a 10GB disk which wasn’t used for anything:

elatov@kerch:~$sudo fdisk -l /dev/sdb

Disk /dev/sdb: 10.7 GB, 10737418240 bytes
255 heads, 63 sectors/track, 1305 cylinders, total 20971520 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x00000000

Disk /dev/sdb doesn't contain a valid partition table

So let’s partition it into two 5GB partitions:

elatov@kerch:~$sudo fdisk /dev/sdb
Device contains neither a valid DOS partition table, nor Sun, SGI or OSF disklabel
Building a new DOS disklabel with disk identifier 0x22f4ab55.
Changes will remain in memory only, until you decide to write them.
After that, of course, the previous content won't be recoverable.

Warning: invalid flag 0x0000 of partition table 4 will be corrected by w(rite)

Command (m for help): n
Partition type:
   p   primary (0 primary, 0 extended, 4 free)
   e   extended
Select (default p): p
Partition number (1-4, default 1): 
Using default value 1
First sector (2048-20971519, default 2048): 
Using default value 2048
Last sector, +sectors or +size{K,M,G} (2048-20971519, default 20971519): +5G 

Command (m for help): n
Partition type:
   p   primary (1 primary, 0 extended, 3 free)
   e   extended
Select (default p): p
Partition number (1-4, default 2): 
Using default value 2
First sector (10487808-20971519, default 10487808): 
Using default value 10487808
Last sector, +sectors or +size{K,M,G} (10487808-20971519, default 20971519): 
Using default value 20971519

Command (m for help): p

Disk /dev/sdb: 10.7 GB, 10737418240 bytes
255 heads, 63 sectors/track, 1305 cylinders, total 20971520 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x22f4ab55

   Device Boot      Start         End      Blocks   Id  System
/dev/sdb1            2048    10487807     5242880   83  Linux
/dev/sdb2        10487808    20971519     5241856   83  Linux

Command (m for help): t
Partition number (1-4): 1
Hex code (type L to list codes): 8e
Changed system type of partition 1 to 8e (Linux LVM)

Command (m for help): t
Partition number (1-4): 2
Hex code (type L to list codes): 8e
Changed system type of partition 2 to 8e (Linux LVM)

Command (m for help): p

Disk /dev/sdb: 10.7 GB, 10737418240 bytes
255 heads, 63 sectors/track, 1305 cylinders, total 20971520 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x22f4ab55

   Device Boot      Start         End      Blocks   Id  System
/dev/sdb1            2048    10487807     5242880   8e  Linux LVM
/dev/sdb2        10487808    20971519     5241856   8e  Linux LVM

Command (m for help): w
The partition table has been altered!

Calling ioctl() to re-read partition table.
Syncing disks.

Now we have two partitions:

elatov@kerch:~$lsblk /dev/sdb
NAME   MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sdb      8:16   0  10G  0 disk 
├─sdb1   8:17   0   5G  0 part 
└─sdb2   8:18   0   5G  0 part 

So let’s add the first one as a physical Volume to LVM:

elatov@kerch:~$sudo pvcreate /dev/sdb1
  Writing physical volume data to disk "/dev/sdb1"
  Physical volume "/dev/sdb1" successfully created

Now let’s add a new volume group:

elatov@kerch:~$sudo vgcreate test /dev/sdb1
  Volume group "test" successfully created
elatov@kerch:~$sudo vgs test
  VG   #PV #LV #SN Attr   VSize VFree
  test   1   0   0 wz--n- 5.00g 5.00g

Now let’s add a new logical volume taking up all the free space:

elatov@kerch:~$sudo lvcreate -l +100%FREE -n lv test
  Logical volume "lv" created
elatov@kerch:~$sudo lvs test
  LV   VG   Attr     LSize Pool Origin Data%  Move Log Copy%  Convert
  lv   test -wi-a--- 5.00g     

So now we have /dev/sdb1 used for the test-lv LV:

elatov@kerch:~$lsblk /dev/sdb
NAME               MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sdb                  8:16   0  10G  0 disk 
├─sdb1               8:17   0   5G  0 part 
│ └─test-lv (dm-2) 254:2    0   5G  0 lvm  
└─sdb2               8:18   0   5G  0 part 

So let’s put ext4 on the LV:

elatov@kerch:~$sudo mkfs.ext4 /dev/mapper/test-lv 
mke2fs 1.42.5 (29-Jul-2012)
Filesystem label=
OS type: Linux
Block size=4096 (log=2)
Fragment size=4096 (log=2)
Stride=0 blocks, Stripe width=0 blocks
327680 inodes, 1309696 blocks
65484 blocks (5.00%) reserved for the super user
First data block=0
Maximum filesystem blocks=1342177280
40 block groups
32768 blocks per group, 32768 fragments per group
8192 inodes per group
Superblock backups stored on blocks: 
	32768, 98304, 163840, 229376, 294912, 819200, 884736

Allocating group tables: done                            
Writing inode tables: done                            
Creating journal (32768 blocks): done
Writing superblocks and filesystem accounting information: done 

Now let’s mount it under /test:

elatov@kerch:~$sudo mkdir /test
elatov@kerch:~$sudo mount /dev/mapper/test-lv /test

elatov@kerch:~$df -h -t ext4
Filesystem              Size  Used Avail Use% Mounted on
/dev/mapper/test-lv     5.0G  138M  4.6G   3% /test

Create a Backup of the LVM metadata

Before we go any further let’s make a backup:

elatov@kerch:~$sudo vgcfgbackup -f /etc/lvm/backup/mybackup test
  Volume group "test" successfully backed up.

You can check out available backups but running vgcfgrestore:

elatov@kerch:~$sudo vgcfgrestore -l test

  File:		/etc/lvm/archive/test_00000-1866804748.vg
  VG name:    	test
  Description:	Created *before* executing 'vgcreate test /dev/sdb1'
  Backup Time:	Sun Aug 10 08:47:36 2014


  File:		/etc/lvm/archive/test_00001-1148580986.vg
  VG name:    	test
  Description:	Created *before* executing 'lvcreate -l +100%FREE -n lv test'
  Backup Time:	Sun Aug 10 08:50:52 2014


  File:		/etc/lvm/backup/mybackup
  VG name:    	test
  Description:	Created *after* executing 'vgcfgbackup test'
  Backup Time:	Sun Aug 10 08:55:10 2014

Notice there are archives and backups. By default, before any lvm command is executed a backup is created:

elatov@kerch:~$grep 'archive =' /etc/lvm/lvm.conf -B 3
	# Should we maintain an archive of old metadata configurations.
	# Use 1 for Yes; 0 for No.
	# On by default.  Think very hard before turning this off.
	archive = 1

So we didn’t really have to create our own, but instead of checking all the archives we know exactly which one we want to back up to. You can also check out the contents of the backup if you desire:

elatov@kerch:~$sudo cat /etc/lvm/backup/mybackup
# Generated by LVM2 version 2.02.95(2) (2012-03-06): Sun Aug 10 10:09:59 2014

contents = "Text Format Volume Group"
version = 1

description = "vgcfgbackup -f /etc/lvm/backup/mybackup test"

creation_host = "kerch"	# Linux kerch 3.2.0-4-amd64 #1 SMP Debian 3.2.60-1+deb7u3 x86_64
creation_time = 1407686999	# Sun Aug 10 10:09:59 2014

test {
	id = "gQ7NxF-uSH4-NsDQ-oWi2-jc2J-Ky6Z-TMy5n6"
	seqno = 3
	format = "lvm2" # informational
	status = ["RESIZEABLE", "READ", "WRITE"]
	flags = []
	extent_size = 8192		# 4 Megabytes
	max_lv = 0
	max_pv = 0
	metadata_copies = 0

	physical_volumes {

		pv0 {
			id = "JIrYxo-FzX5-5oMa-qwO7-SPnQ-noZ2-K4WEY5"
			device = "/dev/sdb1"	# Hint only

			status = ["ALLOCATABLE"]
			flags = []
			dev_size = 10485760	# 5 Gigabytes
			pe_start = 2048
			pe_count = 1279	# 4.99609 Gigabytes
		}
	}

	logical_volumes {

		lv {
			id = "VUA3s7-4sqU-ujuX-uHeH-CBYa-VTgk-rah3lr"
			status = ["READ", "WRITE", "VISIBLE"]
			flags = []
			creation_host = "kerch"
			creation_time = 1407682252	# 2014-08-10 08:50:52 -0600
			segment_count = 1

			segment1 {
				start_extent = 0
				extent_count = 1279	# 4.99609 Gigabytes

				type = "striped"
				stripe_count = 1	# linear

				stripes = [
					"pv0", 0
				]
			}
		}
	}
}

We can see all the Volume information (size, extents….etc).

Increase the size of the Logical Volume

So let’s say at this point you need some extra space. So let’s add the second partition to our test logical volume:

elatov@kerch:~$sudo pvcreate /dev/sdb2
  Writing physical volume data to disk "/dev/sdb2"
  Physical volume "/dev/sdb2" successfully created
elatov@kerch:~$sudo vgextend test /dev/sdb2
  Volume group "test" successfully extended

Now let’s extend the Logical Volume:

elatov@kerch:~$sudo lvextend -l +100%FREE /dev/mapper/test-lv
  Extending logical volume lv to 9.99 GiB
  Logical volume lv successfully resized

Lastly let’s resize the filesystem to match the new size:

elatov@kerch:~$sudo resize2fs -p /dev/mapper/test-lv 
resize2fs 1.42.5 (29-Jul-2012)
Filesystem at /dev/mapper/test-lv is mounted on /test; on-line resizing required
old_desc_blocks = 1, new_desc_blocks = 1
Performing an on-line resize of /dev/mapper/test-lv to 2619392 (4k) blocks.
The filesystem on /dev/mapper/test-lv is now 2619392 blocks long.

elatov@kerch:~$df -h -t ext4
Filesystem              Size  Used Avail Use% Mounted on
/dev/mapper/test-lv     9.9G  140M  9.2G   2% /test

That all looks good.

Decrease the size of the Logical Volume

Now let’s say you don’t need the extra space and you wanted to revert the above changes. First we need to shrink the filesystem and online filesystem shrinking is not supported. So let’s unmount the filesystem and then resize the filesystem:

elatov@kerch:~$sudo umount /test
elatov@kerch:~$sudo e2fsck -f /dev/mapper/test-lv 
e2fsck 1.42.5 (29-Jul-2012)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
/dev/mapper/test-lv: 11/655360 files (0.0% non-contiguous), 76783/2619392 blocks

elatov@kerch:~$sudo resize2fs -p /dev/mapper/test-lv 4G
resize2fs 1.42.5 (29-Jul-2012)
Resizing the filesystem on /dev/mapper/test-lv to 1048576 (4k) blocks.
Begin pass 3 (max = 80)
Scanning inode table          XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
The filesystem on /dev/mapper/test-lv is now 1048576 blocks long.

elatov@kerch:~$sudo e2fsck -f /dev/mapper/test-lv 
e2fsck 1.42.5 (29-Jul-2012)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
/dev/mapper/test-lv: 11/262144 files (0.0% non-contiguous), 51790/1048576 blocks

Now in the previous post, at this point I did an lvreduce to change the size and I had to figure out the extent size. But now let’s do an LVM restore. So prior to the restore here is how the LV looked like:

elatov@kerch:~$sudo lvdisplay -m test
  --- Logical volume ---
  LV Path                /dev/test/lv
  LV Name                lv
  VG Name                test
  LV UUID                VUA3s7-4sqU-ujuX-uHeH-CBYa-VTgk-rah3lr
  LV Write Access        read/write
  LV Creation host, time kerch, 2014-08-10 08:50:52 -0600
  LV Status              available
  # open                 0
  LV Size                9.99 GiB
  Current LE             2558
  Segments               2
  Allocation             inherit
  Read ahead sectors     auto
  - currently set to     256
  Block device           254:2

  --- Segments ---
  Logical extent 0 to 1278:
	Type		linear
	Physical volume	/dev/sdb1
	Physical extents	0 to 1278

  Logical extent 1279 to 2557:
	Type		linear
	Physical volume	/dev/sdb2
	Physical extents	0 to 1278

We can see there are two segments, now for the restore. First let’s remove the volume group (I wanted to disable the volume group with vgchange -an but it still didn’t help):

elatov@kerch:~$sudo vgchange -an test
  0 logical volume(s) in volume group "test" now active

We can see that 0 volumes are active. If you do a restore, it will be reverted:

elatov@kerch:~$sudo vgcfgrestore -f /etc/lvm/backup/mybackup test
  Restored volume group test
elatov@kerch:~$sudo vgs
  WARNING: Inconsistent metadata found for VG test - updating to use version 6
  VG    #PV #LV #SN Attr   VSize  VFree
  test    2   1   0 wz--n-  9.99g    0 

I found a forum (which is now unavailable, but the same content should be seen here on this, here is a description of the issue:

Root Cause

  • LVM2 metadata is stored in a circular buffer. Older metadata may exist in the buffer if the newer metadata does not fill the buffer.
  • If you restore metadata with a lower seqno and it doesn’t overwrite existing metadata in the buffer with a higher sequence number, the higher-sequence number metadata will be in-tact and readable.
  • vgchange finds the newest sequence number possible and uses that version of metadata when activating a volume group. If the restored metadata is not the newest sequence number, an older sequence number may be used.
  • If you’ve created a new LVM Physical Volume (PV) and there is no existing metadata on one or more volumes, those won’t have the higher sequence number metadata, only the restored metadata which is older, and “Inconsistent Metadata” will be reported.

And here is the fix for it:

Resolution

  • Update seqno in the LVM backup file to a number higher one-higher than was reported by vgchange:
  • For example, if the error messages says “updating to use version 149″, change seqno to 150:

    $ grep seqno backupvg_modified.vg
    seqno = 150
    

Rather than messing with the back up file, I removed the volume group completely:

elatov@kerch:~$sudo vgremove test
Do you really want to remove volume group "test" containing 1 logical volumes? [y/n]: y
  Logical volume "lv" successfully removed
  Volume group "test" successfully removed

Then looking over the backups:

elatov@kerch:~$sudo vgcfgrestore -l test | head -20

  File:		/etc/lvm/archive/test_00000-1866804748.vg
  VG name:    	test
  Description:	Created *before* executing 'vgcreate test /dev/sdb1'
  Backup Time:	Sun Aug 10 08:47:36 2014


  File:		/etc/lvm/archive/test_00001-1148580986.vg
  VG name:    	test
  Description:	Created *before* executing 'lvcreate -l +100%FREE -n lv test'
  Backup Time:	Sun Aug 10 08:50:52 2014


  File:		/etc/lvm/archive/test_00002-1407279216.vg
  VG name:    	test
  Description:	Created *before* executing 'vgextend test /dev/sdb2'
  Backup Time:	Sun Aug 10 09:02:11 2014

I want the one right before the extend command:

elatov@kerch:~$sudo vgs test
  Volume group "test" not found
elatov@kerch:~$sudo vgcfgrestore -f /etc/lvm/archive/test_00002-1407279216.vg test
  Restored volume group test
elatov@kerch:~$sudo vgs test
  VG   #PV #LV #SN Attr   VSize VFree
  test   1   1   0 wz--n- 5.00g    0 

Or I could’ve just done it from the backup that we created (either way works):

elatov@kerch:~$sudo vgcfgrestore -f /etc/lvm/test/mybackup
	  Restored volume group test

We can see that it’s 5GB. Lastly to check the segments on the LV:

elatov@kerch:~$sudo lvdisplay -m test
  --- Logical volume ---
  LV Path                /dev/test/lv
  LV Name                lv
  VG Name                test
  LV UUID                VUA3s7-4sqU-ujuX-uHeH-CBYa-VTgk-rah3lr
  LV Write Access        read/write
  LV Creation host, time kerch, 2014-08-10 08:50:52 -0600
  LV Status              available
  # open                 0
  LV Size                5.00 GiB
  Current LE             1279
  Segments               1
  Allocation             inherit
  Read ahead sectors     auto
  - currently set to     256
  Block device           254:2

  --- Segments ---
  Logical extent 0 to 1278:
	Type		linear
	Physical volume	/dev/sdb1
	Physical extents	0 to 1278

We can see that the /dev/sdb2 segment is no longer there. Now to mount the LV:

elatov@kerch:~$df -h -t ext4
Filesystem              Size  Used Avail Use% Mounted on
/dev/mapper/test-lv     4.0G  138M  3.7G   4% /test

We can see the filesystem is still 4G, this is because I resized it to be a little bit smaller than the original. To get the filesystem back to the original size, just run resize2fs without specifying any size and it will automatically resize it to the partition size:

elatov@kerch:~$sudo resize2fs -p /dev/mapper/test-lv 
resize2fs 1.42.5 (29-Jul-2012)
Filesystem at /dev/mapper/test-lv is mounted on /test; on-line resizing required
old_desc_blocks = 1, new_desc_blocks = 1
Performing an on-line resize of /dev/mapper/test-lv to 1309696 (4k) blocks.
The filesystem on /dev/mapper/test-lv is now 1309696 blocks long.

elatov@kerch:~$df -h -t ext4
Filesystem              Size  Used Avail Use% Mounted on
/dev/mapper/test-lv     5.0G  138M  4.6G   3% /test

For good measure do one more fsck on it:

elatov@kerch:~$sudo umount /test
elatov@kerch:~$sudo e2fsck -f /dev/mapper/test-lv 
e2fsck 1.42.5 (29-Jul-2012)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
/dev/mapper/test-lv: 11/327680 files (0.0% non-contiguous), 55902/1309696 blocks

Modify the Backup file to Restore without Removing the Volume Group

So let’s try the fix that is described in the forum. Let’s figure out which sequence number we need to be higher than:

elatov@kerch:~$sudo vgcfgrestore -f /etc/lvm/backup/mybackup test
  Restored volume group test
elatov@kerch:~$sudo vgs
  WARNING: Inconsistent metadata found for VG test - updating to use version 11
  VG    #PV #LV #SN Attr   VSize  VFree
  kerch   1   2   0 wz--n- 15.76g    0 
  test    2   0   0 wz--n-  9.99g 9.99g

Looks like 11 is the one. Now let’s find the line we need to modify:

elatov@kerch:~$sudo grep seqno /etc/lvm/backup/mybackup
	seqno = 3

Let’s change that setting to be 12:

elatov@kerch:~$sudo grep seqno /etc/lvm/backup/mybackup
	seqno = 12

and now let’s try the restore (don’t forget to disable the volume group with vgchange -an test, before applying the backup):

elatov@kerch:~$sudo vgcfgrestore -f /etc/lvm/backup/mybackup test
  Restored volume group test
elatov@kerch:~$sudo vgs
  WARNING: Inconsistent metadata found for VG test - updating to use version 13
  Removing PV /dev/sdb2 (CxEJ9H-XKMs-jnBd-8qeY-PNYW-h4sb-eQKjMJ) that no longer belongs to VG test
  VG    #PV #LV #SN Attr   VSize  VFree
  kerch   1   2   0 wz--n- 15.76g    0 
  test    1   1   0 wz--n-  5.00g    0 

Notice this time around it actually applied the backup and it removed the disk that we added to add the new space.


Published by Karim Elatov

10 August 2014

Tags