Here are some useful one-liner recipes to help manage Linux Device Mapper (sometimes also known as ‘software RAID’) devices. These are too short to warrant separate recipes. They're here because I need them for myself: no matter how many times I've done this, I always check again.

Create MD With Devices Initially Missing

mdadm --create /dev/md0 --level=1 --raid-devices=2 /dev/sdb3 missing

That is, add the missing keyword in place of any and all missing parts of the array.

Change /dev/md0 to the RAID device name you wish to create, and /dev/sdb3 to the partition you wish to add to the newly created array. This will obviously be razed in the process.

Re-Insert a Previously Removed Device

Sometimes a disk is dropped from an array even though it's healthy. There are good reasons for this, such as a dead host adaptor or a failed power supply on an external disk enclosure. When connectivity is restored, or should you wish to force the disk to be re-inserted, say this:

mdadm --manage /dev/md0 --re-add /dev/sda3

The disk must have previously been part of the array, of course. This is preferable to the next recipe, which registers the disk as an altogether new device.

Hot-add New Devices to an MD Array

mdadm --manage /dev/md0 --add /dev/sda3

In this way, a RAID array may be assembled piecemeal, and the system allowed to synchronise it.

View Status Detail on an MD Array

mdadm --misc -D /dev/md0

Displays essential information on the RAID array in question.

cat /proc/mdstats

Displays less information in a more compact form (but includes more rebuild details).