Prepare Lab
Create a file and use it as a block device
The size of the file is 500 blocks of 4Mb (block size) 4M * 500 ~= 2Gb
dd if=/dev/zero of=fileDevice bs=4M count=500
# 1500+0 records in
# 500+0 records out
# 2097152000 bytes (2.1 GB, 2.0 GiB) copied, 3.4276 s, 612 MB/s
With losetup you can attach a loop device to the fileDevice regular file or block device
Before mounting the file we need to check that there is a free /dev/loopX loopback device that we can use to represent our new block device.
Print the name of the first unused loop device
losetup -f
Attach
losetup /dev/loop0 fileDevice
Check status
losetup
# NAME SIZELIMIT OFFSET AUTOCLEAR RO BACK-FILE DIO LOG-SEC
# /dev/loop0 0 0 0 0 <path>/fileDevice 0 512
Physical Volume
Create Physical Volume
Create a 2GB LVM2 Physical Volume.
pvcreate /dev/loop0
Check Physical Volume
pvs - Display information about physical volumes
pvs
# PV VG Fmt Attr PSize PFree
# /dev/loop0 lvm2 --- 1.95g 1.95g
pvdisplay - Display various attributes of physical volume(s)
pvdisplay
# "/dev/loop0" is a new physical volume of "1.95 GiB"
# --- NEW Physical volume ---
# PV Name /dev/loop0
# VG Name
# PV Size 1.95 GiB
# Allocatable NO
# PE Size 0
# Total PE 0
# Free PE 0
# Allocated PE 0
# PV UUID Ddjibt-s5lv-6ouy-44Xx-vgFN-4zzO-15NKV3
Group Volume
Create Group Volume
Create Group Volume with Pysical Extents (PE) of 4Mb
vgcreate VG0 /dev/loop0
# Volume group "VG0" successfully created
Check Group Volume
vgs - Display information about volume groups
vgs
# VG #PV #LV #SN Attr VSize VFree
# VG0 1 0 0 wz--n- <1.95g <1.95g
vgdisplay - Display volume group information
vgdisplay
# --- Volume group ---
# VG Name VG0
# System ID
# Format lvm2
# Metadata Areas 1
# Metadata Sequence No 1
# VG Access read/write
# VG Status resizable
# MAX LV 0
# Cur LV 0
# Open LV 0
# Max PV 0
# Cur PV 1
# Act PV 1
# VG Size <1.95 GiB
# PE Size 4.00 MiB
# Total PE 499
# Alloc PE / Size 0 / 0
# Free PE / Size 499 / <1.95 GiB
# VG UUID Jtijus-SDrW-p8FF-DMb2-scdk-JfGp-wnhLYl
Logical Volume
Create Logical Volume
lvcreate -n <name> -l <NUMBER OF PE> <volume>
lvcreate -n LV0 -l 50 VG0
# Logical volume "LV0" created.
lvcreate -n <name> -L <size> <volume>
lvcreate -n LV0 -L 200 VG0
# Logical volume "LV0" created.
Check Logical Volume
The LVM functionality is mostly handled by the device mapper that provides virtual block device and redirects any access to these virtual devices to another low-level device.
Check the new created device.
ls -l /dev/mapper/VG0-LV0
# lrwxrwxrwx 1 root root 7 Sep 7 18:28 /dev/mapper/VG0-LV0 -> ../dm-3
ls -l /dev/VG0/LV0
# lrwxrwxrwx 1 root root 7 Sep 7 18:52 /dev/VG0/LV0 -> ../dm-3
lvs - Display information about logical volumes
lvs
# LV VG Attr LSize Pool Origin Data% Meta% Move Log Cpy%Sync Convert
# LV0 VG0 -wi-a----- 200.00m
lvdisplay - Display information about a logical volume
lvdisplay
# --- Logical volume ---
# LV Path /dev/VG0/LV0
# LV Name LV0
# VG Name VG0
# LV UUID LyKSfY-tsyY-KUDY-d6lo-g3eG-hhbj-7BclBv
# LV Write Access read/write
# LV Creation host, time bekaar, 2020-09-07 18:28:09 +0200
# LV Status available
# # open 0
# LV Size 200.00 MiB
# Current LE 50
# Segments 1
# Allocation inherit
# Read ahead sectors auto
# - currently set to 256
# Block device 254:3
Remove Logical Volume
you can remove easily a logical volume
umount /dev/mapper/VG0-LV0
lvremove /dev/mapper/VG0-LV0
# Do you really want to remove active logical volume VG0/LV0? [y/n]: y
# Logical volume "LV0" successfully removed
Create a FileSystem
mkfs.ext4 /dev/VG0/LV0
mount /dev/VG0/LV0 /mnt/lvm
Just to make some tests copy some files inside
cd /mnt/lvm
mkdir backup
cp -rf /etc/* .
Snapshot of Logical Volume
A snapshot is a logical volume which allows you to access to the device in a particluar moment in time (the moment in which the snapshot is created). A snapshot memorizes the changes that will occurs on the original device from the creation moment. The size of the snapstot will only depend on the changes on the origian device. Snapshot only stores the changes and is a copy-on-write (COW) operation.
To take a snapshot the syntax is identical to a create except for the -s.
lvm lvcreate --size <size> --snapshot --name <snap name> <dev>
The size of the snapshot is the max size of changes on the original device
First let’s see out device mapper table using dmsetup
dmsetup table
# VG0-LV0: 0 409600 linear 7:0 2048
Now create a snapshot
lvcreate -s -n SNAP1 -L 200M /dev/VG0/LV0
# Logical volume "SNAP1" created.
This will create a new device
ls -l /dev/VG0/SNAP1
# lrwxrwxrwx 1 root root 7 set 8 11:18 /dev/VG0/SNAP1 -> ../dm-3
Lets' control our device mapper table now
dmsetup table
# VG0-LV0-real: 0 409600 linear 7:0 2048
# VG0-SNAP1-cow: 0 409600 linear 7:0 411648
# VG0-LV0: 0 409600 snapshot-origin 254:4
# VG0-SNAP1: 0 409600 snapshot 254:4 254:5 P 8
and
ls -l /dev/mapper/
# lrwxrwxrwx 1 root root 7 set 8 11:18 VG0-LV0 -> ../dm-0
# lrwxrwxrwx 1 root root 7 set 8 11:18 VG0-LV0-real -> ../dm-1
# lrwxrwxrwx 1 root root 7 set 8 11:18 VG0-SNAP1 -> ../dm-3
# lrwxrwxrwx 1 root root 7 set 8 11:18 VG0-SNAP1-cow -> ../dm-2
As you can notice now there are 4 devices now 2 as espected.
The creation of a snapshot creates another layer of mapping.
Two wrapper LVs are created for the snapshot, VG0-LV0-real and VG0-SNAP1-cow.
Let’s check why this happens.
The existing device for the original LV (VG0-LV0) is no longer of type linear, but instead of type snapshot-origin.
It refers to the device VG0-LV0-real.
The new device VG0-SNAP1 is of type snapshot and refers to the device VG0-LV0-real, too, and also to VG0-LV0-cow.
When a write request is done on the original logical volume (VG0-LV0) two operations will occur:
- The data before to be changed is written on the snapshot [COW] (from VG0-LV0-real to vg0-snap1-cow)
- The data is written on the logical partition (VG0-LV0)
When a read request is done on the snapshot logical volume (VG0-SNAP1) only one of this two operations will occur:
- The data is not changed so the read reqauest id forwarded to the original logical volume (VG0-SNAP1 to VG0-LV0-real)
- The data is changed so the data is read from the snapshot (VG0-SNAP1-cow)
Test Snapshot
Let’s remove some file from the lvm mounted filesystem
rm /mnt/lvm/backup/pam.conf
Mount snaphot
mkdir /mnt/lvm_snaphot
mount /dev/VG0/SNAP1 /mnt/lvm_snaphot
The snapshot size shows in the COW-table size.
lvdisplay /dev/VG0/LV0
# --- Logical volume ---
# LV Path /dev/VG0/LV0
# LV Name LV0
# VG Name VG0
# LV Creation host, time istiophorus, 2020-09-08 11:04:26 +0200
# LV snapshot status source of
# SNAP1 [active]
# LV Status available
# LV Size 200,00 MiB
# Current LE 50
lvdisplay /dev/VG0/SNAP1
# --- Logical volume ---
# LV Path /dev/VG0/SNAP1
# LV Name SNAP1
# VG Name VG0
# LV Creation host, time istiophorus, 2020-09-08 11:18:39 +0200
# LV snapshot status active destination for LV0
# LV Status available
# LV Size 200,00 MiB
# Current LE 50
# COW-table size 200,00 MiB
# COW-table LE 50
# Allocated to snapshot 0,04%
# Snapshot chunk size 4,00 KiB
If you keep deleting/changing files in the original LVM FS we can control how SNAP1 changes
# Allocated to snapshot 0,04%
You can also check
lvs -o +devices
# LV VG Attr LSize Pool Origin Data% Meta% Move Log Cpy%Sync Convert Devices
# LV0 VG0 owi-aos--- 200,00m /dev/loop0(0)
# SNAP1 VG0 swi-aos--- 200,00m LV0 0,04 /dev/loop0(50)
Revert Snapshot
With both the Logical Volumes now unmounted we use the lvconvert command to merge the snapshot with the origin.
You make use the –b option to background the process.
Remenber that tThe snapshot will removed at the end of the process.
lvconvert --merge /dev/VG0/SNAP1
# Merging of volume VG0/SNAP1 started.
# VG0/LV0: Merged: 25,71%
# ...
# VG0/LV0: Merged: 95,81%
# VG0/LV0: Merged: 100,00%
Backup
dd if=/dev/VG0/SNAP1 | gzip > backup.gz
Notice that the size of the backup will be equal to the size of orginal data, not the max size of the snapshot.
Once the backup had been done and saved we can remove the snapshot
lvm lvremove /dev/VG0/SNAP1
What is LVM snapshot is full
If the size of the changes made to the LVM files grow in size the snaphot volume became 100% full and the snapshopt will be signed as corrupted.
lvdisplay /dev/VG0/SNAP1
# LV snapshot status INACTIVE destination for LV0
The mapper table will show you that snapshot is marked as invalid
dmsetup status
the only thing you can do is remove the snapshot
lvremove -f /dev/mapper/VG0-SNAP1
to avoid this situation just remeber to check the snapshot size and eventually exatend the logical volume
lvextend -L +3G /dev/mapper/VG0-SNAP1
Take care of inserting the sign + before the size. This is to indicate to enlarge the acual size of 3Gb and not to set the total size to 3Gb.