2

In an attempt to enlarge raid1 array on a working machine, I am testing following scenario:

  • remove one of two disks
  • add bigger disk, resync
  • add another bigger disk, resync, grow to raid-devices=3
  • "fail" and remove the original, small disk
  • "grow" the device back to raid-devices=2
  • make a backup raid device using the small disk
  • enlarge the main device (now containing two big disks)

Now everything seems to work except one important thing: I am unable to assemble a new md device using the disk "failed" and removed from the original md10 device:

mdadm --assemble /dev/md11 /dev/sde3
mdadm: Found some drive for an array that is already active: /dev/md10
mdadm: giving up

Is there any way to somehow change the identification of the array on such a removed device?

4

mdadm is complaining that you're trying to assemble an array that has a UUID matching one that is already running. You'll need to change the UUID on the /dev/sde3 superblock as you assemble it using the --update=uuid and --uuid=<newuuid> parameters. This should tell mdadm to use the UUID you're passing in when comparing to the running arrays.

To find the old UUID, use mdadm --examine /dev/sde3 and grab the old UUID. You could change it to be unique, or generate a new one.

Your new command would be: mdadm --assemble /dev/md11 --update=uuid --uuid=<newuuid> /dev/sde3

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.