Identifying and removing failed mirrored raid slice

Question

Level 1

45 points

Identifying and removing failed mirrored raid slice

Hi,

I've had a disk fail on a mirrored RAID set. I've already added a new disk, and Disk utility currently shows a three slice raid, one of which indicates as failed.

Presuming that I knew which disk failed, I turned the disk off, and disconnected it. Disk utility promptly went into a rebuild, and shows the removed disk as missing. (I may have disconnected one of the good drives!?)

Question 1: How do I physically identify which disk has failed? There doesn't seem to be any information that is useful in the Info tool in Disk Utility.

Question 2: How do I remove the disk slice so it doesn't appear missing? It apparently isn't just a matter of disconnecting.

Question 3: Should I have to rebuild with the two new disks?

Question 4: Does it matter that the new slice is a partition on a larger disk? (320G disk mirrored with a 320G partition of a 1T disk)

Thanks in advance.

Jeff

2.4G Intel Core 2, Mac OS X (10.6.7)

Posted on Mar 29, 2011 10:06 PM

Reply

Answer 1

dechamp

Level 4

3,495 points

Mar 29, 2011 10:50 PM in response to Jeff Schuh

My Raid drives are not managed by my Mac. They have their own processors and OS management tools.

I'm curious about your use of the term "mirrored". It usually means a complete duplicate - a mirror of another drive (Raid 1).

Striped Arrays (Raid 0), however are not duplicates and spread their information across multiple drives. It can be used to enhance transfer speeds or to create larger volumes that are greater in size than an individual drive, but don't offer any redundancy or data recovery in the event of a drive failure.

It is possible to "Mirror" a Striped array so that the set of drives in one raid are fully backed up on a very large drive, or another striped array.

When you lose a drive in a Mirror, the other drive is OK. When you lose a drive in a Stripe, you lose data.

It sounds like you have actually set up a Raid 5 which uses several disks and distributes parity bits across all of the drives so that any drive can fail and still contain all of your data.

My NAS also is a Raid 5, that uses a 4th drive for redundancy and recovery by using any 3 of the 4 drives. All of the data is still available and inserting a new drive will allow the Raid to rebuilt to a Stripe with redundancy. It works well, but if two drives fail at the same time, you still lose your data.

If your Raid is really Mirrored you would just start using the copy and rebuild the original. If it's not Mirrored, then you do have to identify the bad drive and rebuild with a new one.

Reference for Raid descriptions: http://en.wikipedia.org/wiki/Raid_levels

Reply

Answer 2

travis27

Level 1

1 points

Apr 13, 2011 1:40 PM in response to dechamp

dechamp that answer was quite helpful thank you.

I have a mirrored RAID set (two 1GB drives in the MacPro enclosure) and it just disappeared...the system crashed....and now it's working, but I've got my doubts about its longevity now.

Everything is cloned, so I'm confident of being able to get my data back from a backup, but I'm very interested to know more about this mirror.

I set up a mirror in order to be able to swap out a faulty drive with a new one in the event of a fault.

So if one of the drives fails (and assuming I identify it correctly) what are the steps?

I guess the first step is to remove the faulty drive and replace it with a new formatted drive of the same size.

Next step...? You say copy and rebuild. Do you mean simply copy the data from the good slice to the new drive?

Then what is the next step to get Snow Leopard to recognise that this is the new slice of the raid?

Many thanks for any input.

Reply