Set Partitionable Software RAID-6 on OpenSlug

It works for me, it might even work for you :)

I bet it's same for RAID5? - I am just a show-off!

Discaimer 1: This is a "blind leading the blind" HowTo. I am not a Linux guru. You probably know way more. I just wanted to figure this out and so I did. Please, take these directions with a grain of salt. There are probably more elegant ways of achieving same results. If you know any - do not hesitate to add to this page!

Disclaimer 2: Your configuration is different from mine. Understand what you are doing before you do it. Messing with devices and RAID will destroy your data. Backup everything (and make sure you know how to restore).

The Environment

• My Slug has been de-underclocked, but I do not see how it would possilby have any affect.
• I installed OpenSlug version 2.7 beta. Things may work same on other versions... I do not know.
• I have run turnup init but NOT turnup memstick or its disk/nfs equivalent. Can you do all the same things after a full turnup? Why not? (I haven't tried though.)
• There is a Kensington Dome 7-Port USB 2.0 hub plugged into the "disk" port of my Slug. ( For those who care: the top port is the first one, the rest are numbered left-to-right. Find out more by issuing cat /proc/bus/usb/devices command. For the purpose of RAID this info should be irrelevant.)
• I have plugged six identical usb flash drives into the hub and intend to turn them into RAID6? (it is just a coinsidence that I have 6 devices, although RAID6? makes no sence if you have less than 5 drives.)

Step One: Making Space

This step is only necessary if you haven't turned up your Slug onto and external hard drive or memory device, and are still running with your root on the internal jffs2 file system.

To check use df command. If the output contains among other things something like this:

 Filesystem           1k-blocks      Used Available Use% Mounted on
 /dev/mtdblock4            6528      6036       492  92% /

then your root is still using Slug's internal memory. There isn't much space left there and we need to install a few packages. So I uninstalled a few things I did not expect to be needing. Your choice may be different.

 #ipkg remove lrzsz
 #ipkg remove reiserfsprogs
 #ipkg remove reiserfsprogs-reiserfsck
 #ipkg remove reiserfsprogs-mkreiserfs

That gave me enough space.

Step Two: Installing the Software

 #ipkg update

This will bring your package database up to date.

I strongly recommend installing udev - it will save you from the task of creating devices manually. That's what I did.

 #ipkg install udev
 #ipkg install udev-utils

OpenSlug does not come with RAID functionality, so we need to install it too:

 #ipkg install kernel-module-raid6

If thigs go according to plan, kernel-module-raid6 will cause automatic installation of kernel-module-md@ and kernel-module-xor@@ it depends on.

Now install the management tool:

 #ipkg install mdadm

Last step: let the kernel know about all these wonderful modules:

 #depmod

Some kernel modules invoke depmod automatically during installation, but I had to run it manually. There is no harm in it anyway.

Step Three: Creating Partitions

fdisk -l will show you the drives you got. Figure out which one is which, how many you intend to use, and whether you need to create new partitions. On flash memory the "physica" parameters of the "disk" (heads, sectors, cyliners) do not really mean much. Just make sure that the numbers add up Ok. Otherwise your filesystems will get very confused later.

This is not an fdisk HowTo - read the man page, it's not that bad.

Within fdisk menu use p command and you will get something like

 Disk /dev/sda: 65 MB, 65536000 bytes
 8 heads, 32 sectors/track, 500 cylinders
 Units = cylinders of 256 * 512 = 131072 bytes

    Device Boot      Start         End      Blocks   Id  System
 /dev/sda1               1         500       63984   83  Linux 

In this case the math is simple: 131072 bytes/cylinder * 500 cylinders = 6553600 bytes. If you see any messages about physical and logical parameters of the partition being different - take your time and sort it out. (I ignored it first time around at my own peril)

Create a partition for your raid (it does not usually make much sence to have multiple partitions of the same drive withing the same array).

Use t command within fdisk interface to set partition type to fd - Linux raid autodetect.

Exit fdisk using w commant - that's when all your changes are actually written to the drive.

Rinse and repeat for every component of your future array.

Now on with the fun.

Step Four: Creating RAID

OpenSlug kernel does NOT load RAID module automatically. (Correct me if I am wrong!) In the long run you should probably put something into your /etc/rc.boot or some other appropriate script. (I am soooo not qualified to advice on this - anyone with expertese, please pitch in!)

For now let's do it manually:

 #lsmod

 Module                  Size  Used by
 ixp425_eth             15364  0
 ixp400                615208  1 ixp425_eth

 #modprobe raid6

 #lsmod

 Module                  Size  Used by
 raid6                  90896  1
 md                     31508  3 raid6
 xor                     4008  1 raid6
 ixp425_eth             15364  0
 ixp400                615208  1 ixp425_eth

You guessed it! lsmod lists currently loaded modules and thus gives you a nice "before" and "after" picture. raid6 module should automatically load md and xor modules.

mdadm is a great (and simple) tool for managing RAID. If you haven't dealt with it before, take advantage of the following excellent sources (I would have never succeeded in this adventure without them):

EnableRAIDSupport HowTo

mdadm: A New Tool For Linux Software RAID Management by Derek Vadala

mdadm(8) page on man-wiki

If you have been running RAID and some of your devices have been a part of it, it would halp to run mdadm --zer0-superblock /dev/sdxx to clear old RAID info.

I created a partitionable array with two partitions (see the --auto=p2 option in the command below. Many people would advocate a non-partitioned RAID with LVM on top of it. Makes sence in most cases. But it is a personal choice for responsible adults :-D If you do not want to mess with partitionable RAID just change the option to --auto=md - that will take care of creating the associated block device.

Now, let's create the array:

 #mdadm --create --verbose /dev/raid6iv --level=6 --chunk=32 \
 --raid-devices=6 --auto=p2 /dev/sda1 /dev/sdb1 /dev/sdc1  \
 /dev/sdd1 /dev/sde1 /dev/sdf1

mdadm: layout defaults to left-symmetric mdadm: /dev/sdb1 appears to contain an ext2fs file system

    size=63872K  mtime=Fri Apr  7 23:22:40 2006

mdadm: /dev/sdd1 appears to contain an ext2fs file system

    size=58608K  mtime=Tue Jan  3 06:58:10 2006

mdadm: size set to 63872K Continue creating array? y mdadm: array /dev/raid6iv started.

Since you have followed the links above and enthusiastically absorbed every detail ;-) very little comment is needed: clearly I reused a couple of partitions and mdadm gave me a chance to back off before it wipes them out.

You can see that I named my device /dev/raid6iv instead of traditional /dev/md_d0. Why? - Becuse I could. AND because standard names for partitionable RAID devices get rather cryptic. And because I knew that those cryptic devices would still be created automaticall anyway and will be there for me if I ever want to use them to access my RAID array.

Are we done yet? - Almost.

Let's confirm that our RAID is alive and kicking:

 # cat /proc/mdstat

 Personalities : [raid6]
 md_d0 : active raid6 sdf1[5] sde1[4] sdd1[3] sdc1[2] sdb1[1] sda1[0]
      255488 blocks level 6, 32k chunk, algorithm 2 [6/6] [UUUUUU]
      [==========>..........]  resync = 53.9% (35232/63872) finish=0.5min 

It is syncing the drives now, but will be done soon.

Step Five: Finishing Up

Use fdisk to set up the partitions on your RAID.

 #fdisk /dev/raid6iv1
 #fdisk /dev/raid6iv2

The device names associated with the raid partitions on your system will be different from mine, but they are created automatically, no thinking required :-)

In case of flash memory the "physical" parameters of the device are likely to be WRONG. And so if you accept the default value you may end up with a partition bigger than the array. That's bad. So go to Step Three above, and use the check shown to confirm that the numbers do add up. Then create the partitions.

Format them with the filesystem of your choice. For ext2/ext3 there are important performance considerations covered here

 # mke2fs -j -b 4096 -R stride=8 /dev/raid6iv1
 mke2fs 1.38 (30-Jun-2005)
 Filesystem label=
 OS type: Linux
 Block size=4096 (log=2)
 Fragment size=4096 (log=2)
 49408 inodes, 49404 blocks
 2470 blocks (5.00%) reserved for the super user
 First data block=0
 2 block groups
 32768 blocks per group, 32768 fragments per group
 24704 inodes per group
 Superblock backups stored on blocks:    32768

 Writing inode tables: done
 Creating journal (4096 blocks): done
 Writing superblocks and filesystem accounting information: done

 This filesystem will be automatically checked every 38 mounts or
 180 days, whichever comes first.  Use tune2fs -c or -i to override.

You are DONE! Congratulations! Go ahead, mount your filesystem wherever you want, live long and prosper.

Part Six: Housekeeping Considerations

• Many people believe that it is a good practice to have your RAID described in /etc/mdadm.conf file. Some disagree. Your choice.
• On OpenSlug 2.7beta RAID does NOT start itself. You have to make sure that the kernel modules are loaded and the array is assembled (use mdadm --assemble command with the appropriate options. I do not know which of the initialization scripts is appropriate for this. (Comments? I personally would love to put OpenDebianSlug? on this RAID... Suggestions?)
• I have read somewhere that is recommended to actually stop your RAID as a part of the shutdown process. Maybe even unload the kernel modules? - I do not know, and thus do not date to advise.

Figure it out and add to this document. Good luck!