Mastering Linux Storage (Part 2): Live Disk Expansion & Navigating Sector Boundaries in fdisk

In Part 1 of this series, we covered how to diagnose disk space using df and du, unpacked the structural hierarchy of block devices, and used fdisk to uncover a classic storage bottleneck: a swap partition (/dev/vda2) physically blocking our main root partition (/dev/vda1) from expanding into newly allocated disk sectors.

In this second installment, we will walk through the exact hands-on procedure to expand the underlying hypervisor disk, force the Linux kernel to recognize new space on the fly, and safely rewrite partition sector boundaries in fdisk without losing a single byte of data.

Step 1: Resizing the Virtual Disk Image

Before your guest OS can see more disk space, you must expand the virtual disk image on the host hypervisor. If you are using QEMU/KVM with Virtual Machine Manager (virt-manager), first locate your VM’s .qcow2 file path under disk settings (typically /var/lib/libvirt/images/).

Run the following command on your host machine terminal to append additional storage (for example, +50GB):

Bash

sudo qemu-img resize /var/lib/libvirt/images/Kali.qcow2 +50G

Plaintext

Image resized.

If you check fdisk inside the running guest VM right away, you might notice something frustrating: the total sector count hasn’t changed! This happens because the guest kernel caches block device geometries upon boot.

Step 2: Triggering a Live Kernel Rescan

Instead of interrupting your workflow with a full VM reboot, you can force the guest Linux kernel to re-read the block device size dynamically.

Inside your guest VM terminal, execute:

Bash

echo 1 | sudo tee /sys/class/block/vda/device/rescan

Now, verify that the kernel recognizes the new underlying physical capacity:

Bash

sudo fdisk -l /dev/vda

Plaintext

Disk /dev/vda: 90 GiB, 96636764160 bytes, 188743680 sectors

The physical block device /dev/vda now spans 188,743,680 sectors (~90 GiB). However, /dev/vda1 is still locked at sector 79,509,503.

Step 3: Clearing the Obstacle (Removing the Blocking Partition)

As established in Part 1, partitions must be contiguous. Because /dev/vda2 (the swap partition) sits immediately after /dev/vda1, we cannot stretch /dev/vda1 until /dev/vda2 is removed.

First, turn off active swap memory so the kernel releases the device lock:

Bash

sudo swapoff -a

Next, open fdisk targeting the main block device:

Bash

sudo fdisk /dev/vda

Inside the interactive fdisk shell, delete the blocking swap container and the root partition:

1. Type d and select partition 5 (deletes the logical swap partition).
2. Type d and select partition 2 (deletes the extended partition container).
3. Type d and select partition 1 (deletes the primary root partition).

Don’t Panic: Deleting a partition table entry inside fdisk only alters the disk’s index map in memory—it does not erase the underlying ext4 data filesystem written on those sectors.

Step 4: Re-creating the Partition Table Entry

Now we will recreate partition 1 using the exact same starting sector, but extending the end sector out to the full boundary of the disk.

Inside the same fdisk prompt:

1. Type n for a new partition.
2. Type p to make it a Primary partition.
3. Select Partition number 1.
4. First sector: Type 2048 (This must match your original starting sector noted in Part 1 to prevent data corruption!).
5. Last sector: Press Enter to automatically select the maximum available sector (188743679).

The Critical Signature Prompt

During this step, fdisk will detect your existing filesystem and ask a crucial question:

Plaintext

Created a new partition 1 of type 'Linux' and of size 87.9 GiB.
Partition #1 contains a ext4 signature.

Do you want to remove the signature? [Y]es/[N]o:

⚠️ Type N (No)!

If you type Y, fdisk will wipe the ext4 filesystem metadata header, resulting in complete data loss. Typing N preserves your existing data while updating the boundary markers.

Finally, write the updated partition table to disk:

• Type w and press Enter.

Step 5: Informing the Kernel & Resizing ext4

After exiting fdisk, force the kernel to re-read the updated partition table map without rebooting:

Bash

sudo partprobe /dev/vda

Lastly, expand the ext4 filesystem layer to claim the new partition space:

Bash

sudo resize2fs /dev/vda1

Plaintext

resize2fs 1.47.0 (05-Feb-2023)
Filesystem at /dev/vda1 is mounted on /; on-line resizing required
old_desc_blocks = 5, new_desc_blocks = 11
The filesystem on /dev/vda1 is now 23041945 (4k) blocks long.

Verify your new available disk space:

Bash

df -h /

Plaintext

Filesystem      Size  Used Avail Use% Mounted on
/dev/vda1        87G   35G   48G  43% /

What’s Next?

We have successfully expanded our primary partition from 37.9 GB to 87 GB on a live system without losing any data. However, in deleting /dev/vda5, we removed our system’s swap space entirely.

In Part 3, we will finish modernizing our Linux storage architecture by setting up a lightweight, dynamic Swap File inside /dev/vda1 and cleaning up /etc/fstab to keep system boot times lightning fast.

Stay tuned!