Programster's Blog

Tutorials focusing on Linux, programming, and open-source

Create RAID with LVM

Many Linux users have created RAID arrays using mdadm commands and do not realize that you can also create a RAID through LVM.


Installing LVM

You may need to install the LVM packages in order to build these arrays.

sudo apt-get install lvm2  

Partition Drives

If setting up against some physical hard drives, you should create one or more partitions on the drive first.

Creating RAID 0

sudo vgcreate [vg name] /dev/sd[x]1 /dev/sd[x2]1 ...
lvcreate -i[num drives] -I[strip size] -l100%FREE -n[lv name] [vg name]
sudo mkfs.[ext4/xfs] /dev/[vg name]/[lv name]
  • Stripe size needs to be a number of the power 2, starting with 4. E.g. 4, 8, 16, 32, 64. If your data is mostly small text files, then use 4. If you are mostly dealing with media then you may want something larger.
  • If you want to use the xfs filesystem, you may need to install xfprogs with sudo apt-get install xfsprogs -y

Creating this RAID array will remove the ability to remove a drive from the VOLUME group later.

Creating RAID 1 (Mirror)


sudo vgcreate $VG_NAME /dev/sd[x]1 /dev/sd[x]1

sudo lvcreate \
  --mirrors 1 \
  --type raid1 \
  -l 100%FREE \
  --nosync \

sudo mkfs.[ext4/xfs] /dev/$VG_NAME/$LV_NAME

You can also create the volume group by specifying the ID path like /dev/disk/by-id/ata-WDC_WD40EFRX-68WT0N0_WD-XXXXXXXXXXXX-part1 instead of /dev/sd[x]1.

Creating RAID 5 (Parity)


sudo vgcreate $VG_NAME /dev/sd[x]1 /dev/sd[x]1 /dev/sd[x]1

sudo lvcreate \
  --type raid5 \
  -l 100%FREE \
  --nosync \

sudo mkfs.[ext4/xfs] /dev/$VG_NAME/$LV_NAME


If you have any type of raid other than RAID 0, then you can scrub the data every now and then to prevent bitrot.

Check RAID Status

You can check the RAID status with the following command (changing the value of the VG_NAME to your volume group name):

sudo lvs -a -o name,copy_percent,devices $VG_NAME

That will output something similar to:

  LV                   Cpy%Sync Devices                                    
  lvm_raid1            100.00   lvm_raid1_rimage_0(0),lvm_raid1_rimage_1(0)
  [lvm_raid1_rimage_0]          /dev/sdb1(1)                               
  [lvm_raid1_rimage_1]          /dev/sda1(1)                               
  [lvm_raid1_rmeta_0]           /dev/sdb1(0)                               
  [lvm_raid1_rmeta_1]           /dev/sda1(0)

Check that the Cpy%Sync is set to 100.00.

Extra Info

LVM Is Using md Under the Hood

As Felipe Franciosi points out in the comments, configuring as above will still use "md" behind the scenes. It just saves you the trouble of using "mdadm".

You can confirm that by identifying the device mapper setup by lvm with:

dm=$(basename $(readlink /dev/${VG_NAME}/${LV_NAME}))
dmsetup table /dev/${dm}

It will show you that the driver "raid" is being used. Then, from dmsetup(8), you'll see:

"raid Offers an interface to the kernel's software raid driver, md"

For more information on LVM and MD RAID, please refer to this unix & Linux post.

Pysical Migration

Even though RedHat have documentation about how to physically migrate a volume group of disks from one server to another, I found that I could just physically move a RAID 1 LVM pair of disks from one computer to aother and they just showed up without issue. This was from an Xubuntu 20.04 desktop to an Xubuntu 22.04 desktop, which may have made my life easier. I made sure to update both machines /etc/fstab accordingly after the fact.


Last updated: 11th April 2023
First published: 16th August 2018