Resizing a KVM Virtual Disk: A Step-by-Step Guide

Expanding a VM’s storage is a routine operation in KVM environments. The process has two distinct phases: expanding the raw disk capacity and then growing the partitions and filesystems to use that space.

Resizing the Disk Image

The first step is to increase the virtual disk size using qemu-img:

qemu-img resize disk.qcow2 +10G

This adds 10GB to the image. You can also specify an absolute size:

qemu-img resize disk.qcow2 50G

For raw image files, the syntax is identical:

qemu-img resize disk.raw +20G

The VM must be shut down before resizing. If you attempt to resize a disk while the guest is running, qemu-img will refuse with an error unless you use the --shrink flag (which is dangerous and rarely needed).

You can check the current size of an image before resizing:

qemu-img info disk.qcow2

Growing Partitions and Filesystems

After resizing the image, shut down the guest and boot it back up. The new space won’t be available until the guest’s partition table and filesystem are expanded.

Using virt-resize (Recommended)

For qcow2-backed VMs, virt-resize from the libguestfs-tools package handles partition and filesystem expansion in one step without needing to boot the guest:

virt-resize --expand /dev/sda3 disk.qcow2 disk-resized.qcow2

This expands the last partition (/dev/sda3) to fill available space. You can then replace the original image:

mv disk-resized.qcow2 disk.qcow2

virt-resize works offline, making it safer for production VMs. It handles most common filesystem types automatically.

Manual Expansion Inside the Guest

If you prefer to expand partitions manually or are working with LVM/Ceph-backed disks, boot the guest and resize from inside:

For ext4:

sudo fdisk /dev/sda
# Delete the partition, recreate it with larger size, save and exit
sudo resize2fs /dev/sda3

Or use parted for non-destructive resizing:

sudo parted /dev/sda
(parted) resizepart 3 100%
(parted) quit
sudo resize2fs /dev/sda3

For XFS:

sudo parted /dev/sda
(parted) resizepart 3 100%
(parted) quit
sudo xfs_growfs /

XFS can grow while mounted, so you don’t need to unmount.

For LVM:

If your guest uses LVM, expand the physical volume first:

sudo pvresize /dev/sda3
sudo lvextend -l +100%FREE /dev/vg0/lv_root
sudo resize2fs /dev/vg0/lv_root

Storage Backend Considerations

qcow2 files: Use the commands above directly.

Raw images: qemu-img resize works the same way. virt-resize also supports raw images.

LVM-backed disks: Resize the logical volume instead:

sudo lvresize -L +10G /dev/vg0/vm-disk

Then grow the guest’s filesystem as shown above.

Ceph RBD images: Expand the RBD volume:

rbd resize --size 50G pool/image-name

Then resize the guest filesystem.

Verification

After resizing, verify the changes:

df -h
lsblk
parted /dev/sda print

If you resized LVM or Ceph, also check:

pvs
lvs

Watch for any filesystem or partition errors. Most modern filesystems handle online growth well, but always keep backups of important VMs before major storage operations.