Managing Flexible Storage with LVM on Rocky Linux 8 and 9

The Problem: Static Partitioning Rigidity

Standard disk partitioning fixes a specific amount of space to a directory (like /var or /home) at the time of installation. If a log directory fills up or a database grows beyond its initial allocation, you typically face a choice between risky partition moves or adding new physical disks and migrating data manually. This rigidity often leads to system crashes during critical production spikes.

The Takeaway: Logical Volume Management (LVM) abstracts physical storage into a pool of capacity. This allows you to expand volumes on-the-fly, span a single volume across multiple physical disks, and create snapshots for safe system updates without needing to repartition the drive.

Prerequisites

• A system running Rocky Linux 8 or 9.
• Root or sudo privileges.
• At least one unallocated disk or partition (e.g., /dev/sdb).

Implementing the LVM Hierarchy

LVM operates in three layers: Physical Volumes (PV), Volume Groups (VG), and Logical Volumes (LV). You must configure them in this specific order.

Step 1: Initialize Physical Volumes

Prepare the raw disk for LVM use. Run this on the terminal as root:

pvcreate /dev/sdb

Risk: This command initializes the disk; any existing data on /dev/sdb will be inaccessible.

Step 2: Create a Volume Group

A Volume Group acts as a storage pool. You can combine multiple PVs into one VG.

vgcreate vg_data /dev/sdb

Here, vg_data is the name assigned to the pool.

Step 3: Provision Logical Volumes

Carve a usable volume from the pool. For example, to create a 20GB volume for application data:

lvcreate -L 20G -n lv_appdata vg_data

The resulting device will be located at /dev/vg_data/lv_appdata.

Step 4: Format and Mount

LVM creates the virtual disk, but you still need a filesystem. Rocky Linux defaults to XFS.

mkfs.xfs /dev/vg_data/lv_appdata
mkdir /mnt/appdata
mount /dev/vg_data/lv_appdata /mnt/appdata

Expanding Storage Online

The primary advantage of LVM is the ability to grow a volume while the system is running. If lv_appdata reaches capacity, you can extend it provided there is free space in vg_data.

Example: Adding 10GB to a Volume

1. Extend the Logical Volume:

   lvextend -L +10G /dev/vg_data/lv_appdata

2. Grow the Filesystem: The OS sees the larger block device, but the filesystem must be told to use it. For XFS (the Rocky Linux default), use xfs_growfs:

   xfs_growfs /mnt/appdata

Comparison: XFS vs. EXT4 in LVM

Feature	XFS (Default)	EXT4
Online Expansion	Supported	Supported
Online Shrinking	Not Supported	Supported (Offline)
Performance	Better for large files/volumes	Better for small files/volumes

Verification and Diagnostics

To verify the current state of your storage hierarchy, use these three diagnostic commands:

• pvs: Checks physical volume health and total size.
• vgs: Shows how much free space remains in the pool for future expansions.
• lvs: Lists all logical volumes and their current allocation.

Finally, run df -h /mnt/appdata to confirm the filesystem recognizes the new capacity.

Limitations and Rollback

Limitations: LVM adds a small layer of overhead to I/O. More importantly, XFS filesystems cannot be shrunk. If you allocate 100GB to an XFS volume, you cannot reduce it to 50GB later; you would have to back up the data, delete the LV, recreate it smaller, and restore the data.

Rollback: If a volume creation was a mistake, remove the layers in reverse order:

1. Unmount the volume: umount /mnt/appdata
2. Remove the LV: lvremove /dev/vg_data/lv_appdata
3. Remove the VG: vgremove vg_data
4. Remove the PV: pvremove /dev/sdb