RAID info

How to setup RAID

• mdadm wiki
• Some older slackware specific info
• slackware 12 raid setup
• tldp raid info
• RAID

1. First, you must fdisk the drives that you would like to have in your array. Lets assume we're using /dev/hde and /dev/hdg - I create a primary partition on each disk and set its type to “fd” (Linux Raid Autodetect)
2. Check to see if you have any md devices already (we use the next available device in the next step):
   

   # ls -l /dev/md*
   drwxr-xr-x 2 root root 80 2008-02-16 21:07 /dev/md
   lrwxrwxrwx 1 root root  4 2008-02-16 21:07 /dev/md0 -> md/0

   I already have a /dev/md0 device, so I'll use /dev/md1 in the next step.

3. now we create the RAID devices:
   

   # mdadm --create /dev/md1 --level=1 --raid-devices=2 /dev/hd[eg]1

4. I also have a disk, /dev/hdf, a 250gb drive which currently has data on it. However, I want to raid this disk, so I'll be creating another RAID device using a spare 250gb drive (/dev/hdh). I'll mark the second disk as missing, copy the data onto the new 250gb RAID device, and then add /dev/hdf to the RAID device (thus replacing the missing disk).

   1. # mdadm --create /dev/md2 --level=1 --raid-devices=2 /dev/hdh1 missing

   2. # pvcreate /dev/md2
      # vgextend raidvg0 /dev/md2
      # lvextend -l +59651 /dev/raidvg0/wspace_media (extends by the number of "+extents")

   3. now to resize my partition:
      

      # resize2fs /dev/raidvg0/wspace_media

5. Now to add the second drive to the /dev/md2 device:
   1. set the partition type to fd (Linux Raid Autodetect)
   2. add the new device:
      

      # mdadm --add /dev/md2 /dev/hdf1

1. Setup the disks (I have one disk missing from the array):

   root@artemis:/home/john# fdisk /dev/sda (sdb, sdc, sdd)
   
   The number of cylinders for this disk is set to 91201.
   There is nothing wrong with that, but this is larger than 1024,
   and could in certain setups cause problems with:
   1) software that runs at boot time (e.g., old versions of LILO)
   2) booting and partitioning software from other OSs
      (e.g., DOS FDISK, OS/2 FDISK)
   
   Command (m for help): p
   
   Disk /dev/sda: 750.1 GB, 750156374016 bytes
   255 heads, 63 sectors/track, 91201 cylinders
   Units = cylinders of 16065 * 512 = 8225280 bytes
   
      Device Boot      Start         End      Blocks   Id  System
   /dev/sda1               1       91201   732572001   fd  Linux raid autodetect
   
   Command (m for help): 

2. Create the RAID 5 Array (check filesystem/raid tweaking for fine-tuning):

   root@artemis:/home/john# mdadm --create /dev/md4 --level=5 --raid-devices=4 /dev/sd[abc]1 missing
   root@artemis:/home/john# mdadm --detail /dev/md4
   /dev/md4:
           Version : 00.90.03
     Creation Time : Sun Jul 27 19:13:23 2008
        Raid Level : raid5
        Array Size : 2197715712 (2095.91 GiB 2250.46 GB)
     Used Dev Size : 732571904 (698.64 GiB 750.15 GB)
      Raid Devices : 4
     Total Devices : 3
   Preferred Minor : 4
       Persistence : Superblock is persistent
   
       Update Time : Sun Jul 27 19:13:23 2008
             State : clean, degraded
    Active Devices : 3
   Working Devices : 3
    Failed Devices : 0
     Spare Devices : 0
   
            Layout : left-symmetric
        Chunk Size : 64K
   
              UUID : 3190d1da:092e295e:959943bc:2f00e666
            Events : 0.1
   
       Number   Major   Minor   RaidDevice State
          0       8        1        0      active sync   /dev/sda1
          1       8       17        1      active sync   /dev/sdb1
          2       8       33        2      active sync   /dev/sdc1
          3       0        0        3      removed

3. Save the RAID info (not required):

   root@artemis:/home/john# mdadm --detail --scan >> /etc/mdadm.conf

4. Create the filesystem:

   root@artemis:/home/john# mkfs.ext3 /dev/md4

From Ferg's Gaff: Growing a RAID 5 Array:

mdadm --add /dev/md1 /dev/sdf1
mdadm --grow /dev/md1 --raid-devices=4

This then took about 3 hours to reshape the array.

The filesystem the needs to be expanded to fill up the new space.

fsck.ext3 /dev/md1
resize2fs /dev/md1

mdadm wiki page on growing RAID arrays

• Fail all the devices, then remove them, then stop the raid. eg

  # mdadm --manage /dev/mdfoo --fail /dev/sdfoo
  # mdadm --manage /dev/mdfoo --remove /dev/sdfoo
  # mdadm --manage --stop /dev/mdfoo

• http://prefetch.net/articles/linuxsoftwareraid.html

You can do a lot of tweaking by changing the stripe/cluster size and block size.

• stripe/chunk size: size of each stripe block (RAID level) - a 64KB stripe size means that each block of each stripe is 64KB
  • apparently a smaller stripe size is better for fewer large files, and
  • large stripe sizes are better for larger numbers of small files
  • To check stripe/chunk size, run:

    # mdadm --detail /dev/md4

• block size: size of each block (individual disk level) - this should be less than or equal to the stripe/chunk size?
  • not entirely sure how this relates to the stripe/chunk size yet…
  • To check block size, run:

    # dumpe2fs -h /dev/device#

http://www.pcguide.com/ref/hdd/perf/raid/concepts/perfStripe-c.html

http://www.pcguide.com/ref/hdd/perf/raid/levels/single_Level0.htm