Based on this article and a few similar others I see how to simulate a mdadm failure, but it seems it's going to require a rebuild of my array.

I don't want to give it that much IO and also it took days for me to add a disk to the array last time.

Can anybody confirm if the following requires a rebuild of array, or if mdadm will just come back online?

mdadm -manage -set-faulty /dev/md0 /dev/sda1
mdadm /dev/md0 -r /dev/sda1
mdadm /dev/md0 -a /dev/sdc2

"This could be a degraded RAID or perhaps a system in the middle of a reconstruction process. We wait until recovery ends before setting things back to normal." -- http://tldp.org/HOWTO/Software-RAID-HOWTO-6.html


1 Answer 1


You remove a disk from the array; the array needs rebuilding when it's re-added. The rebuild is automatic, but can take time.

However the rebuild time can be minimised if you have a write intent bitmap set on the volume. If the number of changes are small enough then when the disk is re-added it will effectively do a recovery by bringing the "stale" disk back up to date. This can change a rebuild from being hours down to seconds.

See https://raid.wiki.kernel.org/index.php/Write-intent_bitmap for details.

  • Seems like my best option, thank you, I will test and accept! Commented Jul 25, 2016 at 19:32

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .