creating replacing and resizing mdadm Raid 1

Raid controller

Sometimes hardware Raid controllers are very expensive to fit in low budget solutions, when this happens you can use software raid, like mdadm.

Normally hard disks have a limited lifetime. To avoid disasters I replace hard disk every two years, at same time that means that new hard disk will come in a bigger capacity. I will simulate this workflow.

Unfortunately I’m not a rich engineer yet and need to simulate hard disk, if you want to donate some hard disk leave me a comment 馃檪

step 1 and 2 are for create virtual hard disk,

REMEMBER MAKE BACKUPS BEFORE DO DANGEROUS THINGS LIKE THESE

Step 1: Create virtual hard disk

Imagine that in a beginning we have a couple of two hard disk with a capacity of one terabyte (1HD and 2HD), and after two years聽 we acquired two new hard drives with a capacity of three terabyte (3HD and 4HD).

I will use gigabytes instead terabytes for time and capacity reasons

as root

  • cd
  • mkdir mdadmtesting
  • cd mdadmtesting
  • fallocate -l 1G 1HD.raw
  • fallocate -l 1G 2HD.raw
  • fallocate -l 3G 3HD.raw
  • fallocate -l 3G 4HD.raw

Step 2: Associate created files with loop devices

  • losetup /dev/loop0 /root/mdadmtesting/1HD.raw
  • losetup /dev/loop1 /root/mdadmtesting/2HD.raw
  • losetup /dev/loop2 /root/mdadmtesting/3HD.raw
  • losetup /dev/loop3 /root/mdadmtesting/4HD.raw

Step3: create a mdadm device in raid 1 mode

  • 聽mdadm –create –verbose /dev/md0 –level=1 –raid-devices=2 /dev/loop0 /dev/loop1
  • mkfs.ext4 /dev/md0

finally we need to mount new raid and make some files

  • mount /dev/md0 /mnt
  • cd /mnt
  • for i in {1..100}; do echo $i > $i; done

Step 4 Replace one hard disk

we need to mark one hard disk as fail to replace

  • mdadm –manage /dev/md0 –fail /dev/loop0

remove disk from array

  • mdadm –manage /dev/md0 –remove /dev/loop0

attach new hard disk

  • mdadm –manage /dev/md0 –add /dev/loop2

check raid status and wait until raid state is setted as clear and not

State : clean, degraded, recovering

  • mdadm –detail /dev/md0

Step 5: Replace second hard disk

  • mdadm –manage /dev/md0 –fail /dev/loop1
  • mdadm –manage /dev/md0 –remove /dev/loop1
  • mdadm –manage /dev/md0 –add /dev/loop3

wait until state is clean

  • mdadm –detail /dev/md0

Step 6 Grow raid device

At this moment we have removed our old 1Tb disk and they have been replaced with a new 3Tb drive but our raid size is 1Tb we need to grow it

  • mdadm –grow /dev/md0 –size=max

Step 7 Grow filesystem

Our raid size is 3Tb but our file system is still at 1Tb we need to resize it

  • resize2fs /dev/md0

and finally we have our mdadm raid with new drives and a lot of free space.

Enjoy

One Reply to “creating replacing and resizing mdadm Raid 1”

  1. I will like to know if you can replicate this on my server.

    I won’t mind having a chat with you.

    Because a lot of people are against this process

Leave a Reply

Your email address will not be published. Required fields are marked *