← Back to Course Index

Week 4 - Lab: KVM Storage Management

Module: Operating Systems 3 (Virtualisation & Cloud Technologies)

Topic: KVM Disk Images, LVM, Snapshots, and NFS
Estimated Time: 120 Minutes

Lab Overview

In this lab, you will go "under the hood" of virtualization storage. Instead of relying on the Proxmox GUI, you will work directly with the KVM storage layer. You will create raw disk images, manage them with qemu-img, partition them with LVM, and perform manual snapshots and backups. Finally, you will configure networked storage using NFS.

Objectives:

1. Create and manage Virtual Disk Images (raw and qcow2) using qemu-img.
2. Manage storage inside a Linux VM using Logical Volume Manager (LVM).
3. Perform internal snapshots using KVM command-line tools.
4. Execute "Cold Backups" by manipulating disk image files directly.
5. Configure Network File System (NFS) for shared storage.

Prerequisites: - A running Proxmox VE host (Shell Access Required). - A Linux VM (for LVM and NFS sections). - Basic understanding of Linux file permissions.

Part 1: Virtual Disk Management (KVM)

At the lowest level, a Virtual Machine's "hard drive" is just a file on the host operating system. We will use the qemu-img tool to manipulate these files.

Step 1: Inspecting Disk Images

1. Open the Proxmox Shell (Datacenter > Node > Shell).

2. Navigate to the image directory: bash cd /var/lib/vz/images/ ls -lh You might see directories for your existing VMs (e.g., 100, 101).

3. Inspect an existing image (replace [VMID] with an actual ID): bash qemu-img info [VMID]/vm-[VMID]-disk-0.qcow2 (If you don't have a VM, skip to Step 2).

4. Field Report:

   • Image Format: [ qcow2 / raw ]
   • Virtual Size: [ _____________________________ ]
   • Disk Size (actual space used): [ _____________________________ ]

Critical Thinking:

• Why is "disk size" often smaller than "virtual size"?
• [ _________________________________________________________________________ ] (Hint: Sparse files)

Step 2: Creating New Images

1. Create a RAW image (1GB): bash qemu-img create -f raw test-disk.raw 1G

2. Create a QCOW2 image (1GB): bash qemu-img create -f qcow2 test-disk.qcow2 1G

3. Compare sizes: bash ls -lh test-disk.*

4. Field Report:

   • RAW file size: [ _____________________________ ]
   • QCOW2 file size: [ _____________________________ ]

Step 3: Format Conversion & Compression

You can convert disk formats (e.g., to save space or import from another hypervisor).

1. Convert RAW to Compressed QCOW2: bash qemu-img convert -c -f raw -O qcow2 test-disk.raw test-disk-compressed.qcow2

2. Verify details: bash qemu-img info test-disk-compressed.qcow2

3. Field Report:

   • Final size: [ _____________________________ ]

Step 4: Resizing Images

1. Expand the QCOW2 image: bash qemu-img resize test-disk.qcow2 +2G

2. Verify new size: bash qemu-img info test-disk.qcow2 Expected: Virtual size should now be 3GB.

Part 2: Managing Storage inside the Guest (LVM)

Now that you understand the "physical" disk (the file on the host), let's look at how the Guest OS manages that space using Logical Volume Manager (LVM).

Note: For this section, use your Ubuntu/Linux VM. You can use the test-disk.raw we created earlier by attaching it to your VM via Proxmox GUI (Hardware > Add > Hard Disk > Import/Use existing), OR simpler: just create a dummy file inside the VM to practice LVM.

Method: Using a loopback file inside the VM (Safe Mode)

1. Log into your Linux VM.

2. Create a 1GB file (simulates a hard drive): bash dd if=/dev/zero of=/root/virtual_disk.img bs=1M count=1000

3. Attach it to a loop device: bash losetup /dev/loop0 /root/virtual_disk.img

4. Initialize Physical Volume (PV): bash pvcreate /dev/loop0

5. Create Volume Group (VG): bash vgcreate lab_vg /dev/loop0

6. Create Logical Volume (LV) (500MB): bash lvcreate -n lab_vol -L 500M lab_vg

7. Format and Mount: bash mkfs.ext4 /dev/lab_vg/lab_vol mkdir /mnt/test_volume mount /dev/lab_vg/lab_vol /mnt/test_volume

8. Verify: bash df -h /mnt/test_volume lvs

9. Field Report:

   • LV Name: [ _____________________________ ]
   • Filesystem Type: [ ext4 / xfs ]

Part 3: VM Snapshots (KVM CLI)

Proxmox GUI uses these same commands in the background. Let's do it manually on a QCOW2 file.

1. Back on the Proxmox Host Shell.

2. Take a snapshot of your test QCOW2 image: bash qemu-img snapshot -c snap1 test-disk.qcow2 -c stands for Create.

3. List snapshots: bash qemu-img snapshot -l test-disk.qcow2

4. Field Report:

   • Snapshot ID: [ _____ ]
   • Snapshot Name: [ _____ ]

5. Restore a snapshot (Hypothetical command): To apply a snapshot, you would use -a (Apply). bash # qemu-img snapshot -a snap1 test-disk.qcow2

Critical Thinking:

• Can you take a snapshot of a raw format disk?
• [ _________________________________________________________________________ ] (Try it on test-disk.raw and see what happens).

Part 4: VM Backups (Image Operations)

A "Cold Backup" is simply copying the disk image while the VM is off.

1. Ensure the VM/Disk is not in use.

2. Create a backup copy: bash cp test-disk.qcow2 test-disk.qcow2.backup

3. Verify the backup: bash ls -lh test-disk.qcow2*

4. Simulate Corruption: bash echo "CORRUPTION" > test-disk.qcow2 # The file is now broken.

5. Restore from Backup: bash rm test-disk.qcow2 cp test-disk.qcow2.backup test-disk.qcow2

6. Field Report:

   • Why must the VM be offline for a simple cp backup?
   • [ _________________________________________________________________________ ]

Part 5: ZFS Storage (Advanced)

ZFS is the gold standard for enterprise storage. It features self-healing, instant snapshots, and RAID capabilities without needing a hardware controller. We will simulate this using file-backed storage (since we don't have extra physical disks).

Step 1: Create Simulated Disks

1. Create two 1GB files: bash truncate -s 1G /root/disk

2. Verify the status: bash zpool status You should see pool: tank and state: ONLINE with a mirror of disk1 and disk2.