Scope: Logical Volume Management on Debian-based and RHEL-based distributions
Topics: PV, VG, LV, pvcreate, vgcreate, lvcreate, lvextend, resize2fs, xfs_growfs, cloud volume extension.

1. Understand the LVM Model#

LVM places an abstraction layer between physical disks and filesystems:

• Physical Volume (PV): A disk or partition registered with LVM.
• Volume Group (VG): A storage pool built from one or more PVs.
• Logical Volume (LV): A flexible volume carved from a VG, formatted and mounted like a partition.

An LV appears as /dev/<VG_NAME>/<LV_NAME>. Unlike partitions, LVs can be extended while mounted.

2. Create the Physical Volume#

Warning: this destroys existing data on the target. Run as root:

pvcreate /dev/<DEVICE>1

Verify the result:

pvs

Expected output (example):

PV         VG   Fmt  Attr PSize   PFree
/dev/sdb1  appvg lvm2 a--  <50.00g <50.00g

Options:

• /dev/<DEVICE>1: The partition to register. The whole disk (/dev/<DEVICE>) can also serve as a PV.

3. Create the Volume Group#

vgcreate <VG_NAME> /dev/<DEVICE>1

Verify the result:

vgs

Expected output (example):

VG    #PV #LV #SN Attr   VSize   VFree
appvg   1   0   0 wz--n- <50.00g <50.00g

Add another disk to the pool later with vgextend <VG_NAME> /dev/<DEVICE>2.

4. Create the Logical Volume#

Allocate a fixed size:

lvcreate -L 10G -n <LV_NAME> <VG_NAME>

Or consume all remaining space:

lvcreate -l 100%FREE -n <LV_NAME> <VG_NAME>

Verify the result:

lvs

Expected output (example):

LV       VG    Attr       LSize   Pool Origin Data%  Meta%  Log Cpy%Sync Convert
datalv   appvg -wi-a-----  10.00g

Options:

• -L 10G: Absolute size. Use -L +5G to add to an existing LV.
• -l 100%FREE: Size in extents, here all free space in the VG.
• -n <LV_NAME>: Name of the logical volume.

5. Create the Filesystem and Mount#

Format the LV. Use mkfs.ext4 on Debian-based and mkfs.xfs on RHEL-based systems:

mkfs.ext4 /dev/<VG_NAME>/<LV_NAME>
mkdir -p <MOUNT_POINT>
mount /dev/<VG_NAME>/<LV_NAME> <MOUNT_POINT>
findmnt <MOUNT_POINT>

Persist the mount in /etc/fstab by UUID:

UUID=<UUID>  <MOUNT_POINT>  ext4  defaults  0  2

Obtain the UUID with blkid /dev/<VG_NAME>/<LV_NAME> and test the entry with mount -a.

6. Verify the LVM Stack#

Inspect every layer:

pvs
vgs
lvs

Inspect the device tree:

lsblk

Expected output (example):

sdb
└─sdb1                8:17   0   50G  0 lvm
  └─appvg-datalv    253:0    0   10G  0 /srv/data

7. Extend a Logical Volume#

Grow the VG first when it has no free space:

vgextend <VG_NAME> /dev/<DEVICE>2

Extend the LV by a fixed amount or to use all remaining space:

lvextend -L +10G /dev/<VG_NAME>/<LV_NAME>
lvextend -l +100%FREE /dev/<VG_NAME>/<LV_NAME>

Let LVM resize the filesystem in the same step:

lvextend -r -l +100%FREE /dev/<VG_NAME>/<LV_NAME>

Options:

• -L +10G: Adds 10 GiB to the current size. Without + the value is the new absolute size.
• -l +100%FREE: Adds all remaining VG space.
• -r, --resizefs: Grows the filesystem after the LV using the appropriate tool (resize2fs or xfs_growfs).

When not using -r, the filesystem must be grown separately. Use the steps in Part 1 for ext4 or Part 2 for XFS.

8. Extend a Cloud-Attached Volume#

Scenario: the LV sits on a cloud disk that was enlarged in the provider console, for example from 50 GiB to 100 GiB. Extend each layer in order.

Resize the partition first if the PV is a partition rather than the whole disk:

growpart /dev/<DEVICE> 1

Options:

• growpart: Resizes a partition to fill the device. Provided by cloud-guest-utils on Debian-based and cloud-utils-growpart on RHEL-based systems.

Resize the PV to the new device size:

pvresize /dev/<DEVICE>1

Confirm that the VG now reports free space:

vgs

Extend the LV and filesystem in one step:

lvextend -r -l +100%FREE /dev/<VG_NAME>/<LV_NAME>

Verify the result:

df -h <MOUNT_POINT>

If the PV uses the whole disk, growpart is unnecessary: pvresize alone picks up the new device size.

Part 1: Debian-Based Systems#

Examples: Debian, Ubuntu

1. Grow an ext4 Filesystem#

After lvextend without -r, grow ext4 while mounted:

resize2fs /dev/<VG_NAME>/<LV_NAME>

Verify the result:

df -h <MOUNT_POINT>

2. Check an ext4 Filesystem#

Run on the unmounted LV only:

e2fsck -f /dev/<VG_NAME>/<LV_NAME>

ext4 can be shrunk offline after unmounting, but treat shrinking as a last resort and take a backup first.

Part 2: RHEL-Based Systems#

Examples: RHEL, Rocky Linux, AlmaLinux

1. Grow an XFS Filesystem#

After lvextend without -r, grow XFS from its mount point while mounted:

xfs_growfs <MOUNT_POINT>

Verify the result:

df -h <MOUNT_POINT>

2. Check an XFS Filesystem#

Run on the unmounted LV only:

xfs_repair -n /dev/<VG_NAME>/<LV_NAME>

XFS cannot be shrunk. Plan capacity accordingly and prefer extending over pre-allocating large volumes.

Part 3: Important Notes#

LVM Layers#

Extension Commands#

Security Guidelines#

• Leave free space in the VG for snapshots and emergency growth. Consuming 100% of the VG removes these options.
• Never shrink an XFS filesystem; no supported tooling exists for it.
• Take a backup before any size reduction, fsck, or partition change.
• LVM snapshots consume the same VG space and can fill up silently. Monitor them with lvs and do not treat snapshots as backups.
• Run pvcreate, vgcreate, and mkfs only on devices verified with lsblk and blkid.
• Do not run fsck, resize2fs (shrinking), or xfs_repair on mounted filesystems.
• Do not remove a PV with vgreduce while it still holds extents in use. Check with pvdisplay first.
• Use descriptive VG and LV names such as appvg-datalv so df and lvs output stays readable during incidents.
• When scripting, treat a non-empty vgs VFree column as the precondition for lvextend.