How to Extend an mdadm RAID1 to RAID5 on a UEFI Boot System

Last updated: Feb 15, 2025

I've already shown how to create a RAID1 on a UEFI boot with mdadm on a Debian distribution, you can find that wonderful tutorial here. I will now explain how to extend a RAID1 mdadm on a UEFI boot to a RAID5 as it's quite common, over time, to need more disk space.

I'll assume in the following article that an UEFI RAID1, configured with two 20G disks, is already perfectly functional. A third 20G disk will be added to form a RAID5.

And since I'm in a happy mood, I'll show how to do it with or without lvm (yes, yes, don't thank me! Actually, you can do it… 😊).

⚠️First and foremost, as is always the case when it comes to disk configuration, it's essential to make a backup. The data will simply be unrecoverable if the server is rebooted during the (very long!) process of converting from RAID1 to RAID5. You've been warned!⚠️

Does it work?

I carried out a number of tests while writing this article. Indeed, after migrating the root partition (which here includes the /boot/ partition) to RAID5, the system would no longer boot normally and switches to grub rescue mode. What's strange is that the system boots normally when you first enter the UEFI boot menu and manually select one of the EFI boot partitions. I've seen this behavior on VMware ESXi and Proxmox VE. Fortunately, it seems to work outside the virtual machine, in my case I tested it on a Dell PowerEdge T630. If you have an explanation, I'd be glad to hear it!

Illustration of boot issues in virtualized environments (VMware and Proxmox) after migrating a system to RAID5, showing different boot menus, GRUB rescue mode, and EFI boot selection screens. — Boot issues after RAID5 migration on VMware and Proxmox, requiring manual EFI boot selection.

To sum up, always make backups. Ideally test before going into production.

Now that you've been properly warned, let's get on with the tutorial!

Adding the new disk

As explained in the introduction, we start with a working UEFI RAID1.

Our disk configuration looks like this: two disks (sda and sdc) configured in RAID1 and one newly added disk (sdc):

Diagram illustrating the RAID1 partitioning schema with EFI system partitions, ext4 root partitions, and swap partitions on two disks (sda and sdb), along with an additional unallocated 20G disk (sdc).

Convert RAID1 to RAID5

Identify the new disk to avoid deleting an existing one. Here we have our sdc disk with no partition configured, indicating that this is our new disk:

root@host:~# fdisk -l
Disk /dev/sda: 20 GiB, 21474836480 bytes, 41943040 sectors
Disk model: Virtual disk    
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: F89D780E-776F-4CFF-A096-2F13B8435BE4

Device        Start      End  Sectors  Size Type
/dev/sda1      2048    98303    96256   47M EFI System
/dev/sda2     98304 39159807 39061504 18.6G Linux RAID
/dev/sda3  39159808 41940991  2781184  1.3G Linux RAID


Disk /dev/sdb: 20 GiB, 21474836480 bytes, 41943040 sectors
Disk model: Virtual disk    
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: 580CC43C-5BED-45D4-B2B5-FD460A3EC1A1

Device        Start      End  Sectors  Size Type
/dev/sdb1      2048    98303    96256   47M EFI System
/dev/sdb2     98304 39159807 39061504 18.6G Linux RAID
/dev/sdb3  39159808 41940991  2781184  1.3G Linux RAID


Disk /dev/sdc: 20 GiB, 21474836480 bytes, 41943040 sectors
Disk model: Virtual disk    
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes

An alternative way of identifying the new disk is to use the cat /proc/mdstat command. Here we see our RAID arrays for sda and sdb:

root@host:~# cat /proc/mdstat 
Personalities : [raid1] [linear] [multipath] [raid0] [raid6] [raid5] [raid4] [raid10] 
md100 : active raid1 sda1[0] sdb1[2]
      48064 blocks super 1.0 [2/2] [UU]
      
md1 : active (auto-read-only) raid1 sda3[0] sdb3[1]
      1388544 blocks super 1.2 [2/2] [UU]
      
md0 : active raid1 sda2[1] sdb2[0]
      19513344 blocks super 1.2 [2/2] [UU]

Install the gdisk tool:

root@host:~# apt update && apt install gdisk

Copy the partitions scheme from sdb to sdc using gdisk:

root@host:~# sgdisk /dev/sdb -R /dev/sdc

Regenerate a GUID of the sdc disk to avoid having the same as that of sdb copied beforehand:

root@host:~# sgdisk -G /dev/sdc

After these commands, the partitions of our sdc disk should be defined:

Diagram illustrating the transition from RAID1 to RAID5 with the addition of a third disk (sdc), including an EFI partition, an 18G RAID partition, and a 2G RAID swap partition.

Add an EFI entry named debian3 for the EFI partition of our sdc disk:

root@host:~# efibootmgr --create --disk /dev/sdc --part 1 --label "debian3" --loader "\EFI\debian\shimx64.efi"

Add disk sdc1 to RAID1 md100 array. Note that we're not converting our UEFI partition to RAID5, because if we do, the boot system won't be able to read a RAID5 partition:

root@host:~# mdadm --grow /dev/md100 --raid-devices=3 --add /dev/sdc1

Our md100 array is now a RAID1 composed of three partitions:

Diagram showing the RAID1 to RAID5 conversion, now including an EFI partition on the third disk (sdc), with synchronized RAID partitions for system and swap space.

This command converts the array to RAID5 and adds the sdc2 partition to the md0 array (SYSTEM). (Optional, but consider using the --backup-file option in case of a system crash during the operation):

root@host:~# mdadm --grow /dev/md0 --raid-device=3 --level=5 -a /dev/sdc2

Check reshaping progress (⚠️ Do not reboot the system while reshaping is in progress⚠️):

root@host:~# cat /proc/mdstat
cat /proc/mdstat 
Personalities : [raid1] [linear] [multipath] [raid0] [raid6] [raid5] [raid4] [raid10] 
md100 : active raid1 sdc1[3] sdb1[0] sda1[2]
      48064 blocks super 1.0 [3/3] [UUU]
      
md1 : active (auto-read-only) raid1 sda3[0] sdb3[1]
      1388544 blocks super 1.2 [2/2] [UU]
      
md0 : active raid5 sdc2[2] sda2[0] sdb2[1]
      19513344 blocks super 1.2 level 5, 64k chunk, algorithm 2 [3/3] [UUU]
      [>....................]  reshape =  0.3% (77596/19513344) finish=37.5min speed=8621K/sec

Our md0 array is now a RAID5 consisting of three partitions:

Diagram illustrating the final RAID5 configuration, with three disks (sda, sdb, sdc) synchronized, including EFI partitions, RAID5 for the system, and RAID1 for swap.

Do the same with the sdc3 partition on the md1 array (SWAP):

root@host:~# mdadm --grow /dev/md1 --raid-device=3 --level=5 -a /dev/sdc3

Our md1 array is now also a RAID5 with three disks:

Final RAID5 setup with three synchronized disks (sda, sdb, sdc), including EFI partitions, RAID5 for the system partition, and RAID5 for the swap partition.

Resize partitions

Resize the File System

Once shaping is complete, resize the file sytem to the new size (example here with ext4 file system):

root@host:~# resize2fs /dev/md0

Check that the new size has been taken into account:

root@host:~# df -h
Filesystem      Size  Used Avail Use% Mounted on
udev            1.9G     0  1.9G   0% /dev
tmpfs           392M  700K  391M   1% /run
/dev/md0         37G  2.3G   33G   7% /
tmpfs           2.0G     0  2.0G   0% /dev/shm
tmpfs           5.0M     0  5.0M   0% /run/lock
/dev/md100       47M  5.9M   41M  13% /boot/efi
tmpfs           392M     0  392M   0% /run/user/0
tmpfs           392M     0  392M   0% /run/user/1000

Resize the Swap

For the swap partition, we need to disable and then recreate the swap partition. This change involves modifying the entries in /etc/fstab and /etc/mdadm/mdam.conf.

Disable the swap:

root@host:~# swapoff /dev/md1

Create a new swap that takes the new size into account, and copy the new UUID:

root@host:~# mkswap /dev/md1
mkswap: /dev/md1: warning: wiping old swap signature.
Setting up swapspace version 1, size = 2.6 GiB (2843734016 bytes)
no label, UUID=dfefc270-893d-4c77-930e-94d2985c9ab3

Note that we can retrieve the UUID with the blkid command:

root@host:~# blkid /dev/md1
/dev/md1: UUID="dfefc270-893d-4c77-930e-94d2985c9ab3" TYPE="swap"

Edit /etc/fstab and modify the swap entry to include the new UUID:

# /etc/fstab: static file system information.
#
# Use 'blkid' to print the universally unique identifier for a
# device; this may be used with UUID= as a more robust way to name devices
# that works even if disks are added and removed. See fstab(5).
#
# systemd generates mount units based on this file, see systemd.mount(5).
# Please run 'systemctl daemon-reload' after making changes here.
#
# <file system> <mount point>   <type>  <options>       <dump>  <pass>
# / was on /dev/md0 during installation
UUID=869d77af-e5b0-4f17-bfd8-79245b6915d6 /               ext4    errors=remount-ro 0       1
# /boot/efi was on /dev/sda1 during installation
#UUID=4FE8-AC32  /boot/efi       vfat    umask=0077      0       1
#/dev/md100 /boot/efi vfat umask=0077	0	1
/dev/md100 /boot/efi vfat umask=0077,noauto,defaults	0	1
# swap was on /dev/md1 during installation
#OLD SWAP ENTRY: UUID=42f44cc3-025d-4a22-8510-647ff25338de none            swap    sw              0       0
UUID=dfefc270-893d-4c77-930e-94d2985c9ab3 none            swap    sw              0       0
#/dev/sr0        /media/cdrom0   udf,iso9660 user,noauto     0       0

Next, edit the /etc/initramfs-tools/conf.d/resume file and replace the UUID as for the /etc/fstab file:

RESUME=UUID=dfefc270-893d-4c77-930e-94d2985c9ab3

Generate a new initramfs image for the changes to take effect:

root@host:~# update-initramfs -u

Activate swap:

root@host:~# swapon /dev/md1

Check swap size:

root@host:~# swapon --show
NAME     TYPE      SIZE USED PRIO
/dev/md1 partition 2.6G   0B   -2

With LVM

It's pretty much the same thing, the difference being that we need to use the pvresize command to make our LVM take the new space into account. As we can see, LVM brings flexibility.

Let's suppose we're in this LVM configuration:

Diagram illustrating a RAID1 setup with LVM partitions, including EFI partitions, logical volumes for boot, root, usr, var, tmp, home, and swap, on two synchronized disks (sda and sdb), with an additional unallocated disk (sdc).

Copy the partitions scheme from sdb to sdc using gdisk:

root@host:~# sgdisk /dev/sdb -R /dev/sdc

Regenerate a GUID of the sdc disk to avoid having the same as that of sdb copied beforehand:

root@host:~# sgdisk -G /dev/sdc

Add an EFI entry named debian3 for the EFI partition of our sdc disk:

root@host:~# efibootmgr --create --disk /dev/sdc --part 1 --label "debian3" --loader "\EFI\debian\shimx64.efi"

Add partition sdc1 to RAID1 md100 array:

root@host:~# mdadm --grow /dev/md100 --raid-devices=3 --add /dev/sdc1

Add partition sdc2 and convert the array to RAID5:

root@host:~# mdadm --grow /dev/md0 --raid-device=3 --level=5 -a /dev/sdc2

Check reshaping progress (⚠️ Do not reboot the system while reshaping is in progress⚠️):

root@host:~# cat /proc/mdstat

Once fully reshaped, expand the LVM:

root@host:~# pvresize /dev/md0

Resize the file sytem of each logical volume you wish to resize:

root@host:~# lvextend -r -L+2G /dev/mapper/vgos-lvhome