RED HAT ENTERPRISE LINUX

LVM Physical Volumes:
Create and Remove

Create and remove physical volumes

CIS126RH | RHEL System Administration 1
Mesa Community College

Logical Volume Management (LVM) is the standard storage abstraction layer on RHEL. Before logical volumes and filesystems can be created, raw storage — a partition or whole disk — must be initialised as an LVM physical volume (PV). Physical volumes are the foundation of the LVM hierarchy, and knowing how to create, inspect, and remove them is the first required skill in the LVM toolkit. This module is tested on the RHCSA exam.

Learning Objectives

1. Explain the LVM architecture — Describe the three-layer LVM hierarchy: physical volumes, volume groups, and logical volumes, and the role each layer plays
2. Prepare a partition for use as a physical volume — Create a partition with the correct type code and use pvcreate to initialise it as an LVM physical volume
3. Inspect physical volumes — Use pvs, pvdisplay, and pvscan to list and examine physical volume details
4. Remove physical volumes safely — Use pvremove to uninitialise a physical volume and understand the prerequisites for safe removal

The LVM Three-Layer Architecture

LVM adds a flexible abstraction layer between raw storage and filesystems. The three layers build on each other from bottom to top.

1. Physical Volume (PV) — a partition or whole disk initialised for LVM use. Divided internally into fixed-size physical extents (PEs). Commands: pvcreate, pvs, pvremove
2. Volume Group (VG) — one or more PVs pooled together into a single storage pool. The VG is the unit of administration. Commands: vgcreate, vgs, vgremove
3. Logical Volume (LV) — a flexible "virtual partition" carved from VG space. Filesystems are placed on LVs, not on partitions directly. Commands: lvcreate, lvs, lvremove

LVM Concepts: Physical Extents

A physical volume is divided into fixed-size blocks called physical extents (PEs). Extents are the atoms of LVM storage.

• Default PE size is 4 MiB — set when the volume group is created
• Every LVM allocation happens in whole PE increments
• A 10 GiB PV with 4 MiB PEs contains 2560 physical extents
• Free PEs in a VG can be allocated to logical volumes
• When a PV is added to a VG, its PEs join the VG's free extent pool

# pvdisplay shows PE size and count for a physical volume
$ sudo pvdisplay /dev/sdb1
  --- Physical volume ---
  PV Name               /dev/sdb1
  VG Name               datavg
  PV Size               10.00 GiB
  Allocatable           yes
  PE Size               4.00 MiB
  Total PE              2559
  Free PE               1024
  Allocated PE          1535
  PV UUID               abc123...

Preparing Storage for LVM

A physical volume can be created on a partition or directly on a whole disk. Using a partition is the standard practice on most systems.

Option A — Use a partition (recommended)

# Create a partition with the LVM type code (8e00 in gdisk)
$ sudo parted --script /dev/sdb \
    mklabel gpt \
    mkpart primary 1MiB 100%
$ sudo partprobe /dev/sdb
# Then initialise the partition as a PV
$ sudo pvcreate /dev/sdb1

Option B — Use the whole disk (no partition table)

# Initialise the entire disk directly — no partitioning needed
$ sudo pvcreate /dev/sdb
# LVM writes its metadata at the beginning of the disk

Creating a Physical Volume: pvcreate

pvcreate writes LVM metadata to the beginning of a partition or disk, marking it as an LVM physical volume ready to join a volume group.

# Initialise a single partition as a physical volume
$ sudo pvcreate /dev/sdb1
  Physical volume "/dev/sdb1" successfully created.

# Initialise multiple partitions at once
$ sudo pvcreate /dev/sdb1 /dev/sdc1
  Physical volume "/dev/sdb1" successfully created.
  Physical volume "/dev/sdc1" successfully created.

# Initialise a whole disk as a physical volume
$ sudo pvcreate /dev/sdb
  Physical volume "/dev/sdb" successfully created.

# Verify the PV was created
$ sudo pvs
  PV         VG   Fmt  Attr PSize   PFree
  /dev/sda3  rhel lvm2 a--  <17.00g      0
  /dev/sdb1       lvm2 ---  <10.00g <10.00g

Listing Physical Volumes: pvs

pvs (physical volume summary) provides a concise one-line-per-PV listing — the standard first command to check PV status.

# Summary listing of all physical volumes
$ sudo pvs
  PV         VG     Fmt  Attr PSize    PFree
  /dev/sda2  rhel   lvm2 a--   <19.00g       0
  /dev/sdb1         lvm2 ---   <10.00g  <10.00g
  /dev/sdb2  datavg lvm2 a--    5.00g    2.00g

# Show additional columns
$ sudo pvs -o +pv_used,pv_free,pv_pe_count,pv_pe_alloc_count

# List all available output fields
$ sudo pvs -o help

Detailed PV Inspection: pvdisplay

pvdisplay shows complete details for one or all physical volumes — the verbose counterpart to pvs.

# Show details for a specific PV
$ sudo pvdisplay /dev/sdb1
  --- Physical volume ---
  PV Name               /dev/sdb1
  VG Name               datavg
  PV Size               10.00 GiB / not usable 4.00 MiB
  Allocatable           yes
  PE Size               4.00 MiB
  Total PE              2559
  Free PE               512
  Allocated PE          2047
  PV UUID               Abc123-DE45-fg67-HI89-jk01-LM23-NO45pq

# Show all physical volumes
$ sudo pvdisplay

# Short format — one line summary per PV
$ sudo pvdisplay --short
  "/dev/sdb1" is a member of volume group "datavg"
  "/dev/sdb2" is NOT a member of any volume group

pvscan: Discovering Physical Volumes

pvscan scans all block devices for LVM physical volume metadata and reports what it finds — useful for confirming LVM sees a new PV.

# Scan all devices for LVM physical volumes
$ sudo pvscan
  PV /dev/sda2   VG rhel   lvm2 [<19.00 GiB / 0    free]
  PV /dev/sdb1   VG datavg lvm2 [10.00 GiB / 2.00 GiB free]
  PV /dev/sdb2           lvm2 [5.00 GiB]
  Total: 3 [<34.00 GiB] / in use: 2 [<29.00 GiB] / in no VG: 1 [5.00 GiB]

# Scan and cache LVM data (useful after hot-plug)
$ sudo pvscan --cache

# Check if a specific device is a PV
$ sudo pvck /dev/sdb1
  Checking physical volume (metadata area #0) of /dev/sdb1.

What pvcreate Writes

Understanding what pvcreate writes to the disk helps explain why a PV can be uninitialised with pvremove.

• pvcreate writes a small LVM label in one of the first four sectors of the device
• The label contains: a magic signature, the PV UUID, the device size, and a pointer to the metadata area
• The metadata area stores the full LVM configuration including VG membership, LV definitions, and PE allocations
• LVM metadata is stored in a circular text file format — human-readable with pvck --dump
• Removing a PV with pvremove wipes the label and metadata, leaving the underlying data intact

Removing a Physical Volume: pvremove

pvremove uninitialises a physical volume by wiping its LVM label and metadata. The device is then no longer recognised as an LVM PV.

# Remove a physical volume that is NOT in any volume group
$ sudo pvremove /dev/sdb2
  Labels on physical volume "/dev/sdb2" successfully wiped.

# Confirm it is gone from pvs output
$ sudo pvs
  PV         VG     Fmt  Attr PSize   PFree
  /dev/sda2  rhel   lvm2 a--  <19.00g      0
  /dev/sdb1  datavg lvm2 a--  <10.00g  <2.00g
# /dev/sdb2 no longer appears

# Attempting to remove a PV that is in a VG fails
$ sudo pvremove /dev/sdb1
  PV /dev/sdb1 is used by VG datavg so please remove using vgreduce first.

Safe PV Removal Sequence

Removing a PV from an active volume group requires moving its data first to preserve the logical volumes that use it.

1. Confirm the PV has used extents
   sudo pvs /dev/sdb1 — check Allocated PE is non-zero
2. Move all data off the PV
   sudo pvmove /dev/sdb1 — relocates all LEs to other PVs in the VG
3. Remove the PV from the volume group
   sudo vgreduce datavg /dev/sdb1
4. Uninitialise the PV
   sudo pvremove /dev/sdb1
5. Verify the PV is gone
   sudo pvs — confirm /dev/sdb1 no longer appears

Extending a VG by Adding a Physical Volume

One of the most common LVM tasks is adding a new PV to an existing volume group to expand its capacity. This is a non-disruptive online operation.

# Current state — VG is nearly full
$ sudo vgs
  VG     #PV #LV #SN Attr   VSize   VFree
  datavg   1   2   0 wz--n- <10.00g 500.00m

# Step 1: Prepare the new disk/partition
$ sudo parted --script /dev/sdc mklabel gpt mkpart primary 1MiB 100%
$ sudo partprobe /dev/sdc

# Step 2: Initialise as a PV
$ sudo pvcreate /dev/sdc1
  Physical volume "/dev/sdc1" successfully created.

# Step 3: Add to the existing volume group
$ sudo vgextend datavg /dev/sdc1
  Volume group "datavg" successfully extended

# Confirm the VG now has more free space
$ sudo vgs
  VG     #PV #LV #SN Attr   VSize   VFree
  datavg   2   2   0 wz--n- <20.00g <10.50g

Physical Volume Commands Quick Reference

LVM Command Family Overview

LVM commands follow a consistent naming pattern: pv* for physical volumes, vg* for volume groups, lv* for logical volumes.

Common Mistakes

Knowledge Check

Answer these before moving to the next slide.

1. What are the three layers of LVM, from bottom to top? What is each layer's purpose?
2. Write the command to initialise /dev/sdb1 as an LVM physical volume.
3. After running pvcreate, write the command to confirm the PV was created and show its size and free space.
4. The output of pvs shows a PV with an empty VG column and PFree equal to PSize. What does this indicate?
5. You need to remove /dev/sdb1 from the volume group datavg and uninitialise it. It currently has 2 GB of allocated extents. Write the commands in the correct order.
6. What is a physical extent, and why does its size matter?

Knowledge Check — Answers

1. Physical Volume (PV) — raw storage (partition or disk) initialised for LVM. Volume Group (VG) — a pool of one or more PVs; the unit of storage administration. Logical Volume (LV) — a flexible virtual partition carved from VG space; filesystems are placed on LVs.
2. sudo pvcreate /dev/sdb1
3. sudo pvs — the output should show /dev/sdb1 with an empty VG column, its PSize, and PFree equal to PSize (no extents allocated yet). Alternatively, sudo pvdisplay /dev/sdb1 for full details.
4. The PV has been initialised with pvcreate but has not yet been added to any volume group. All of its space is available — none of its physical extents have been allocated to logical volumes.
5. 1. sudo pvmove /dev/sdb1 — move all data off /dev/sdb1 to other PVs in datavg
   2. sudo vgreduce datavg /dev/sdb1 — remove /dev/sdb1 from the volume group
   3. sudo pvremove /dev/sdb1 — wipe the LVM metadata from /dev/sdb1
6. A physical extent (PE) is the smallest unit of storage that LVM allocates. It is typically 4 MiB and is set when the volume group is created. The PE size matters because: all LVM allocations must be in whole PE increments — the minimum logical volume size is one PE; capacity calculations depend on the PE size; and once set for a VG, it cannot be changed without recreating the VG.

Key Takeaways

1. LVM has three layers: PV → VG → LV. Physical volumes are the raw storage foundation. Volume groups pool PVs into a single capacity pool. Logical volumes are flexible virtual partitions carved from the pool. Filesystems go on logical volumes.
2. Create a physical volume with pvcreate /dev/DEVICE. The device must be a partition or whole disk, unmounted, and the kernel must know it exists (run partprobe after partitioning). Verify with pvs or pvdisplay.
3. Inspect with pvs (summary) and pvdisplay (detail). An empty VG column means the PV is not in any volume group. Free PE × PE size = available space. Use pvscan to discover all PVs.
4. Remove safely: pvmove → vgreduce → pvremove. A PV in a VG cannot be directly removed. Move data off with pvmove, remove from the VG with vgreduce, then uninitialise with pvremove. Skipping pvmove causes data loss.

Graded Lab

• Run lsblk to identify an unpartitioned disk in your lab VM. Create a GPT partition table and a single partition using all available space with parted --script. Run partprobe and confirm with lsblk.
• Run sudo pvcreate /dev/sdb1 to initialise the partition as a physical volume. Confirm success with sudo pvs.
• Run sudo pvdisplay /dev/sdb1 and record: PV Size, PE Size, Total PE, Free PE. Calculate the available space in MiB (Free PE × PE Size).
• Run sudo pvscan to see the system-wide PV summary. Note that the new PV is shown as "in no VG."
• Create a second partition on the same disk (/dev/sdb2) using 2 GiB. Initialise it as a PV. Run sudo pvs to confirm both /dev/sdb1 and /dev/sdb2 appear.
• Remove the second PV: run sudo pvremove /dev/sdb2. Confirm it is gone with sudo pvs. Delete partition 2 from the disk with parted and run partprobe.

Next: Getting Help from Local Documentation