Arch Linux

jac

Member

From: /home/jac

Registered: 2009-05-19

Posts: 431

Website

Building RAID, hard drive failure?

I just reinstalled Arch to go with disk encryption and to setup my spare drives (4 2TiB HDDs) as one RAID5. Now, I started the process with

mdadm --create /dev/md0 --level=5 --raid-devices=4 /dev/sd{b,c,d,e}1

After about 11 hours, it was going at half the speed it started out at, and then eventually it failed saying that sdb1 was bad. I've now got this in /proc/mdstat:

Personalities : [raid6] [raid5] [raid4] 
md127 : inactive sdd1[2] sdc1[1] sde1[4]
      5860532931 blocks super 1.2
       
unused devices: <none>

So, two questions:
1) I've checked the drive with hdsentinel, and it says it is fine -- in fact, it get's a 100% in health, while two other drives in the array have less than that. Should I check it with something else as well? Should I just try to add it to the array? Should I get it replaced, just in case?

2) Whenever I have a drive I want to add to this array to complete it (I wanted RAID5 so I would need 4 2TiB drives to get 6TiB of space), how do I add it? I assume the process will be the same no matter if it's the same drive, just trying again, or a brand new one. I couldn't find anything on our wiki about it, and the information I'm finding about failures isn't really that understandable to me. Could someone point me to a good resource for this?

Thanks!

byte

Member

From: Düsseldorf (DE)

Registered: 2006-05-01

Posts: 2,046

Re: Building RAID, hard drive failure?

1) Run 'smartctl -t long' on all the disks, ideally when they're not in use/not mounted. It will tell you when it should be done, and you can also check 'smartctl -c' for percentage of completion. But it will take a while on large drives. Afterwards have a look at 'smartctl -A ... | grep Sector' and 'smartctl -l selftest' - there should be zero or close to zero reallocated/pending sectors and the self-test should be "completed without errors".

2) man mdadm

---

1000

wsims

Member

Registered: 2010-04-15

Posts: 34

Re: Building RAID, hard drive failure?

In this example you would first add the disk to the array, then grow the array.

mdadm --add /dev/md1 /dev/sdd1
mdadm --grow /dev/md1 --raid-devices=4

jac

Member

From: /home/jac

Registered: 2009-05-19

Posts: 431

Website

Re: Building RAID, hard drive failure?

Blah! Well, look at RAW_VALUE (I hope that is the one I want), sdb is fine. However the selftest failed on two of the other drives. I think this may be an issue with my controller, I have to get my motherboard replaced anyway so I will test again after that.

Thank you both for your instructions!