Archive

Archive for March, 2010

Reflections on Google Apps

March 24th, 2010 No comments

An awesome writeup from Paul Hite on migration considerations for Google Apps:

http://www.paulhite.com/2010/03/reflections-on-google-apps.html

Categories: Tech Tags:

2GB/sec from SSD Raid?

March 22nd, 2010 No comments

Interesting video regarding SSDs and RAID.  No technical backup data to verify this, but it does seem feasible!

http://goo.gl/f7Of

Categories: Tech Tags:

Checking a linux MD raid array

March 15th, 2010 1 comment

Sorry for the long delay in posting.  Illness and long work hours prevented much time for this, but I just wanted to offer this little tidbit…

For those of you familiar with Linux, you may be aware that it supports both hardware and software RAID types.  Naturally hardware RAID tends to make better use of the resources available due to things such as bus and CPU offloading, but software RAID typically leaves you with more control should something disastrous happen.

In this little post I’ll go over a quick-and-easy method of verifying the soft-RAID array.  A lot of sysadmins I know will use mdadm to force a rebuild which is…clunky at best.  I believe the example below to be safer.  In my example, my RAID array will be md0.

fs1:/ # cd /sys/block/md0/md

fs1:/sys/block/md0/md # echo check >sync_action
[Wait a few moments]
fs1:/sys/block/md0/md # cat /proc/mdstat
Personalities : [raid6] [raid5] [raid4]
md0 : active raid5 sdc1[0] sde1[3] sdd1[1]
1953503488 blocks super 1.0 level 5, 128k chunk, algorithm 0 [3/3] [UUU]
[>....................]  check =  3.1% (30306176/976751744) finish=283.3min speed=55672K/sec
bitmap: 0/466 pages [0KB], 1024KB chunk
unused devices: <none>

You can check on the /proc/mdstat as often as you like during the consistency check.  The RAID is up and running during all of this and you don’t have the risk of a disk failing while the array is broken [which could happen during the mdadm method].  Not saying that a disk won’t fail, but if it does then at least the RAID is as intact before it failed rather than definitely being broken!
Comments, as always, welcome.
Categories: Commands, Linux, Tech Tags: