Loading...

VONDS ©

Lab 125: LVM Basics

Back to Lab Index

Build an LVM stack from scratch (PV → VG → LV) on a blank disk (/dev/vdb), format the LV with XFS, mount it at /mnt/logs, make the mount persistent with a UUID-based /etc/fstab entry, extend the LV, grow XFS online, then revert the system back to a clean state.

storage lvm xfs rhel fstab

Scenario

A VM gets a new blank disk for application logs. You want flexibility to resize later, so you choose LVM. You will create PV → VG → LV, format XFS, mount at /mnt/logs, make it persistent via /etc/fstab, extend the LV, grow XFS online, then clean up.

Safety note

LVM operations write metadata to disks and can destroy data if you target the wrong device. Confirm the target disk and verify each layer (PV/VG/LV) before formatting or mounting.

Objective

  • Confirm /dev/vdb exists, is unused, and has no mountpoints.
  • Install LVM tooling (lvm2) if required.
  • Create and verify an LVM physical volume (PV) on /dev/vdb.
  • Create a volume group (VG) logs_vg and a logical volume (LV) logs_lv.
  • Create XFS on the LV and mount it at /mnt/logs.
  • Persist the mount using a UUID-based /etc/fstab entry and validate with mount -a.
  • Extend the LV and grow XFS online.
  • Cleanly revert: remove fstab entry, unmount, remove LV/VG/PV labels.

Concepts

  • LVM layering: PV → VG → LV and what each layer represents.
  • Device naming: canonical paths (/dev/<vg>/<lv>) vs mapper paths (/dev/mapper/*).
  • Filesystem vs block resizing: grow the LV first, then grow XFS.
  • Persistent mounts: why UUIDs are safer than device names in /etc/fstab.
  • Ops-safe validation: using mount -a after fstab edits to avoid boot failures.
  • Clean teardown order: unmount first, then remove LV → VG → PV to avoid “in use” errors.

Walkthrough

Step 1: Confirm the disk exists and is unused.
Command 
lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINTS

Verify the target disk is present, the size matches expectation, and it has no filesystem or mountpoint. 

NAME   MAJ:MIN RM SIZE RO TYPE FSTYPE MOUNTPOINTS
vda    253:0    0  11G  0 disk
└─vda1 253:1    0  10G  0 part xfs    /
vdb    253:16   0   1G  0 disk
Step 2: Ensure LVM tools are installed.
Command 
sudo dnf -y install lvm2

If the system already has LVM tooling, the package manager should report nothing to do.

Step 3: Create a physical volume (PV).
Command 
sudo pvcreate /dev/vdb

This writes LVM metadata to the disk and makes it available for a volume group.

Step 4: Verify the PV exists and is free.
Command 
sudo pvs -o pv_name,vg_name,pv_size,pv_free
PV       VG  PSize  PFree
/dev/vdb      1.00g  1.00g
Step 5: Create a volume group (VG).
Command 
sudo vgcreate logs_vg /dev/vdb

A VG pools PV capacity into something you can slice into logical volumes.

Step 6: Create a logical volume (LV).
Command 
sudo lvcreate -L 500M -n logs_lv logs_vg

This creates the block device you will format and mount.

Step 7: Verify the LV path.
Command 
sudo lvs -o lv_name,vg_name,lv_size,lv_path

The canonical device path is /dev/logs_vg/logs_lv (and the mapper path is also valid).

Step 8: Create XFS on the LV.
Command 
sudo mkfs.xfs /dev/logs_vg/logs_lv

This is a destructive write. Only run it after you have verified the LV path.

Step 9: Create the mountpoint and mount it.
Commands 
sudo mkdir -p /mnt/logs
sudo mount /dev/logs_vg/logs_lv /mnt/logs

Mounting is your first end-to-end validation that the storage stack is usable.

Step 10: Verify the mount.
Command 
df -h /mnt/logs

This confirms the filesystem is mounted and reports its size.

Step 11: Capture the UUID for a stable fstab entry.
Command 
sudo blkid /dev/logs_vg/logs_lv

Use the filesystem UUID rather than device paths for persistence.

Step 12: Add a UUID-based fstab entry and validate it safely.
Commands 
# Replace UUID below with your actual value from blkid
echo 'UUID=REPLACE_UUID /mnt/logs xfs defaults 0 0' | sudo tee -a /etc/fstab

# Ops-safe validation (should return no output on success)
sudo mount -a
Ops habit

After any change to /etc/fstab, run mount -a immediately. If you broke the file, you want to find out now, not at the next reboot.

Step 13: Extend the LV and grow XFS online.
Commands 
sudo lvextend -L +200M /dev/logs_vg/logs_lv
sudo xfs_growfs /mnt/logs
df -h /mnt/logs

The LV grows first (block device). Then XFS expands online while mounted. Finish by verifying the new size with df.

Alternative

In real work you may use lvextend -r to resize the filesystem automatically. This lab keeps the two steps explicit so you build the habit of verifying each layer.

Common breakpoints

pvcreate fails: “device is in use”

You targeted a disk that already has signatures, partitions, or is mounted. Re-check lsblk -f and confirm you are operating on the correct device.

mkfs.xfs complains about an existing signature

The LV already has a filesystem signature. On real systems, you would investigate why, and only use mkfs.xfs -f if you are intentionally overwriting it.

mount -a errors after editing /etc/fstab

This usually means a typo in the UUID, mountpoint, or filesystem type. Fix the line immediately. Do not leave a broken /etc/fstab behind.

xfs_growfs shows no change

The filesystem can only grow if the LV grew. Re-check lvs after lvextend and confirm the mountpoint is correct.

Cleanup fails because the filesystem is still mounted

Unmount first. If a process is holding the mount busy, you would locate it (for example with lsof or fuser) before retrying.

Cleanup checklist

The goal is to leave the system in a clean, boot-safe state with no leftover mounts and no LVM metadata on the test disk.

Commands 
# 1) Remove the mount entry
sudo sed -i '/\/mnt\/logs/d' /etc/fstab

# 2) Unmount (only after confirming /mnt/logs is mounted)
sudo umount /mnt/logs

# 3) Remove LVM objects (LV -> VG -> PV)
sudo lvremove -y /dev/logs_vg/logs_lv
sudo vgremove -y logs_vg
sudo pvremove -y /dev/vdb

# 4) Final verification (should show vdb as a raw disk with no PV/VG/LV)
lsblk -f
sudo pvs
sudo vgs
sudo lvs
Final verification signal

pvs/vgs/lvs should not show your test objects, and lsblk -f should show /dev/vdb without an LVM signature.

Reference

  • lsblk: List block devices and their layers.
  • dnf: Install packages on RHEL-based systems.
    • -y: Automatically answer "yes" to prompts.
  • pvcreate: Initialize a block device as an LVM Physical Volume (PV).
  • pvs: Display PVs and free space.
    • -o pv_name,vg_name,pv_size,pv_free: Show PV name, VG membership, total size, and free space.
  • vgcreate: Create a Volume Group (VG) from one or more PVs.
  • vgs: Display VGs and free space.
  • lvcreate: Create a Logical Volume (LV) from a VG.
    • -L: Set size.
    • -n: Set LV name.
  • lvs: Display LVs and their paths.
    • -o lv_name,vg_name,lv_size,lv_path: Show key LV fields including canonical path.
  • mkfs.xfs: Create an XFS filesystem on a block device.
  • mount: Mount filesystems.
    • -a: Mount all entries in /etc/fstab (except noauto).
  • df: Report filesystem space usage.
    • -h: Human-readable units.
  • blkid: Display UUID and filesystem metadata.
  • tee: Write stdin to a file (works well with sudo).
    • -a: Append instead of overwrite.
  • lvextend: Extend an LV (block device) before filesystem growth.
    • -L +SIZE: Increase size by the specified amount.
  • xfs_growfs: Grow an XFS filesystem online.
  • sed: Edit text non-interactively.
    • -i: Edit file in place.
  • umount: Unmount a filesystem.
  • lvremove: Remove an LV.
    • -y: Auto-confirm prompts.
  • vgremove: Remove a VG.
    • -y: Auto-confirm prompts.
  • pvremove: Remove LVM metadata from a PV device.
    • -y: Auto-confirm prompts.
  • lsblk -f: Show filesystem signatures and UUIDs (useful for cleanup verification).