To see the status of any RAID setups: cat /proc/mdstat

To create a linear array (spanning 2 or more partitions)

• create 2 partitions (say /dev/sdb1 and /dev/sdb2)
• create the array: mdadm --create --verbose /dev/md0 --level=linear --raid-devices=2 /dev/sdb1 /dev/sdb2
• create a filesystem: mkfs /dev/md0
• mount it!: mnt /dev/md0 /mnt/raid

RAID1

• To create a RAID1 (mirrored) with 1 backup drive:
  mdadm --create --verbose /dev/md0 --level=raid1 --raid-devices=2 -x 1 /dev/sdb1 /dev/sdb2 /dev/sdb3

What to do when an array drive dies (with a safe, SCA-based system. Like one with a SuperMicro SCA814S backplane)

• pull out the bad drive
• tell the raid to remove it mdadm /dev/md0 --remove /dev/sdc1
• replace the drive
• tell the raid to remove it mdadm /dev/md0 --add /dev/sdc1
• let it rebuild!

Misc

• Show details about an array: mdadm --detail /dev/md0
• 'Stop' an array (tell kernel to release it): mdadm -S /dev/md0
• Dump status of arrays on the system: mdadm --detail --scan (can be copied directly into /etc/mdadm.conf)
• Dump status on for a given array: mdadm /dev/md0
• Dump status about one disk and the array it belongs to: mdadm --examine /dev/sdb1
• Assemble previously set up RAID partitions into a RAID volume: mdadm --assemble /dev/md0 /dev/sdb1 /dev/sdb2
• Add a spare disk mdadm /dev/md0 --add /dev/sdb3
• Remove a spare disk mdadm /dev/md0 --remove /dev/sdb3
• Simulate a drive going faulty: mdadm --manage --set-faulty /dev/md0 /dev/sdb2

-- MattWalsh - 15 Dec 2006