Managing LVM Logical Volumes On Linux

LVM (Logical Volume Management) abstracts physical storage into logical volumes, giving you flexibility to resize filesystems, snapshot data, and manage disk space without downtime. It sits between your physical disks and the filesystems you mount.

The three core LVM concepts are:

Physical Volumes (PVs): Physical disks or disk partitions
Volume Groups (VGs): Collections of PVs that act as a storage pool
Logical Volumes (LVs): Virtual partitions carved from a VG

Creating and Initializing LVM

Start by preparing a disk or partition as a physical volume:

# Initialize a physical volume
pvcreate /dev/sdb1

# Verify the PV was created
pvs
pvdisplay /dev/sdb1

Create a volume group from one or more physical volumes:

# Create a VG named 'storage'
vgcreate storage /dev/sdb1

# Add another PV to the existing VG
vgextend storage /dev/sdc1

# View VG details
vgs
vgdisplay storage

Create logical volumes within the volume group:

# Create a 50GB LV named 'data'
lvcreate -L 50G -n data storage

# Or use percentage of available space
lvcreate -l 80%VG -n backup storage

# List logical volumes
lvs
lvdisplay /dev/storage/data

Format and mount the logical volume:

mkfs.ext4 /dev/storage/data
mkdir -p /mnt/data
mount /dev/storage/data /mnt/data

Resizing Logical Volumes

One of LVM’s major advantages is resizing without unmounting (for most filesystems):

# Extend an LV by 20GB
lvextend -L +20G /dev/storage/data

# Or extend to 150GB total
lvextend -L 150G /dev/storage/data

# Resize the filesystem to match the new LV size
resize2fs /dev/storage/data

# For XFS filesystems (which resize online)
xfs_growfs /mnt/data

To shrink a volume, unmount first and shrink the filesystem before the LV:

umount /mnt/data
e2fsck -f /dev/storage/data
resize2fs /dev/storage/data 40G
lvreduce -L 40G /dev/storage/data
mount /dev/storage/data /mnt/data

Removing and Recovering LVM Components

Remove a logical volume:

umount /dev/storage/data
lvremove /dev/storage/data

Remove a volume group:

vgremove storage

Remove a physical volume:

pvremove /dev/sdb1

Handling Missing Physical Volumes

If a disk fails or becomes unavailable, the volume group enters a degraded state. To recover:

# First, identify what's missing
vgdisplay -v storage

# Remove missing PVs from the VG (use only if the disk is permanently gone)
vgreduce --removemissing storage

# For LVs mirrored across multiple PVs, you may need to force removal
vgreduce --removemissing --force storage

Caution: Before removing missing PVs, verify the disk is actually inaccessible and not just temporarily offline. Data loss can occur if you’re too aggressive with --removemissing.

Snapshots

Create point-in-time snapshots of logical volumes for backups or testing:

# Create a 10GB snapshot of the 'data' volume
lvcreate -L 10G -s -n data_snapshot /dev/storage/data

# The snapshot appears as /dev/storage/data_snapshot
mount /dev/storage/data_snapshot /mnt/snapshot

# Remove the snapshot
umount /mnt/snapshot
lvremove /dev/storage/data_snapshot

Monitoring and Maintenance

Monitor LVM health regularly:

# Check physical volume status
pvs -o pv_name,pv_size,pv_free

# Monitor for any warnings or errors
vgck storage

# Get detailed space usage
vgs -o vg_name,vg_size,vg_free,vg_extent_size

# Check specific LV utilization
df -h /mnt/data

Use lvmdevices (available in modern LVM versions) to manage which devices LVM scans:

# List devices LVM will use
lvmdevices

# Update the device filter if needed
lvmdevices --adddev /dev/sdd1

Practical Considerations

Always take snapshots before major operations
Monitor free space in volume groups—when a VG runs full, add more PVs with vgextend
For production systems, consider mirrored or RAID-backed LVs for redundancy
Use consistent naming conventions (e.g., vg_prod, vg_test)
Keep LVM metadata backups: vgcfgbackup and vgcfgrestore can recover from metadata corruption

LVM is essential infrastructure for most Linux servers. Master these basics and you’ll handle storage management efficiently.

Managing LVM Logical Volumes on Linux