Ticket Options
Question Details
TICKET ARCHIVE -> How to Recover From a Failed Raid Mirror?
mmontgomery - Aug 17, 2005 - 11:46 pm
I have a G5 running 10.4.2. The boot drive is a RAID mirrored set of 160 GB drives. The raid volume will not boot. When boot with the Tiger install disk, and run Disk Utility, the raid volume shows both disks OFFLINE. Disk Utility does not seem to give me the option to do anything, other than maybe erase everything and start over.
So I figure the solution must be some commands using the terminal running diskutil. However, I have a very precarious hold on my data, so I want to make sure I do the right thing with diskutil, so as not to lose my data. I would greatly appreciate some expert advice here!
One thing I did try was to use diskutil removefrom RAID to remove one of the disks from the raid volume. However, I could not figure out how to create a volume on the disk without destroying the contents of the disk, something I definitely do not want to do. So now I may be in worse shape than ever. I have already been working for two days on this problem, and I really need to get my computer up, so please help as soon as you can find time.
Background
--------------
How did I get into this situation? I tried to boot my Mac this morning, and it just blinked with the "no disk" icon. (BTW, it did not 'bong' when it started up.) It had been shut down normally the previous day, no problems, no hangs.
I tried to eject my CD drive (pushing down the mouse button), to put in the Tiger disk, but it would not open. So I reset the PRAM. I tried it once, but it did not help, so I did a deep reset. Now the drive opens when I press the mouse.
I put in the Tiger disk, booted, and went to Disk Utility. The Raid volume was shown Degraded. Normally a degraded disk will mount and boot, but not this time. I ran a hardware test that showed both disks were fine. I tried to rebuild the mirror (using the good disk), but nothing happened.
I figured I needed Diskwarrior to come to the rescue. I booted Diskwarrior. It was able to mount the raid volume somehow, and seemed to repair the directory successfully. I checked the directories, and they looked good. All my data seems to be there.
I tried to boot from the repaired drive, but it just blinks. So I booted from Tiger again, and ran Disk Utility. It now showed the Raid drives, the first one marked Spare Offline, and the second drive marked Offline.
Unfortunately, I don't know how to deal with drives that are marked offline, either with Disk Utility or diskutil, so I start a frantic search on this topic.
I find that apparently several people have run into this kind of problem. It seems if you build your Raid in 10.3, and upgrade to 10.4, which is exactly what I did, sometimes RAID drive 0 gets trashed, as mine did.
There was also an Apple note 151950 which indicates that resetting the PRAM can affect your RAID. Then there was an almost helpful Apple note 106987, that gives detailed instructions on how to rebuild a raid mirror using diskutil. Almost helpful, in that it does not talk about an Offline drive, so nothing really applies to my situation.
I finally became desperate enough to call Apple since my Tiger purchase was less than 90 days ago. I waited on hold for an hour, only to get told that Apple does not support Raid, and that there is no where to escalate the matter, Apple can't help me at all. (Apple case # 51805281). I offered to pay them to help me or recommend a third party to help me, but they just said "sorry".
So now I badly need my computer, but I don't have any hope of getting my data back unless one of you kind people can help me!
Please email me at michael@montgo.us as well as post if you have any advice. Thank you!
So I figure the solution must be some commands using the terminal running diskutil. However, I have a very precarious hold on my data, so I want to make sure I do the right thing with diskutil, so as not to lose my data. I would greatly appreciate some expert advice here!
One thing I did try was to use diskutil removefrom RAID to remove one of the disks from the raid volume. However, I could not figure out how to create a volume on the disk without destroying the contents of the disk, something I definitely do not want to do. So now I may be in worse shape than ever. I have already been working for two days on this problem, and I really need to get my computer up, so please help as soon as you can find time.
Background
--------------
How did I get into this situation? I tried to boot my Mac this morning, and it just blinked with the "no disk" icon. (BTW, it did not 'bong' when it started up.) It had been shut down normally the previous day, no problems, no hangs.
I tried to eject my CD drive (pushing down the mouse button), to put in the Tiger disk, but it would not open. So I reset the PRAM. I tried it once, but it did not help, so I did a deep reset. Now the drive opens when I press the mouse.
I put in the Tiger disk, booted, and went to Disk Utility. The Raid volume was shown Degraded. Normally a degraded disk will mount and boot, but not this time. I ran a hardware test that showed both disks were fine. I tried to rebuild the mirror (using the good disk), but nothing happened.
I figured I needed Diskwarrior to come to the rescue. I booted Diskwarrior. It was able to mount the raid volume somehow, and seemed to repair the directory successfully. I checked the directories, and they looked good. All my data seems to be there.
I tried to boot from the repaired drive, but it just blinks. So I booted from Tiger again, and ran Disk Utility. It now showed the Raid drives, the first one marked Spare Offline, and the second drive marked Offline.
Unfortunately, I don't know how to deal with drives that are marked offline, either with Disk Utility or diskutil, so I start a frantic search on this topic.
I find that apparently several people have run into this kind of problem. It seems if you build your Raid in 10.3, and upgrade to 10.4, which is exactly what I did, sometimes RAID drive 0 gets trashed, as mine did.
There was also an Apple note 151950 which indicates that resetting the PRAM can affect your RAID. Then there was an almost helpful Apple note 106987, that gives detailed instructions on how to rebuild a raid mirror using diskutil. Almost helpful, in that it does not talk about an Offline drive, so nothing really applies to my situation.
I finally became desperate enough to call Apple since my Tiger purchase was less than 90 days ago. I waited on hold for an hour, only to get told that Apple does not support Raid, and that there is no where to escalate the matter, Apple can't help me at all. (Apple case # 51805281). I offered to pay them to help me or recommend a third party to help me, but they just said "sorry".
So now I badly need my computer, but I don't have any hope of getting my data back unless one of you kind people can help me!
Please email me at michael@montgo.us as well as post if you have any advice. Thank you!
MacIT - Aug 19, 2005 - 12:58 pm
Michael,
I know how stressful dealing with RAID failures can be. Let's see if we can't move this forward some.
First off, my thinking here is that we want to destroy the RAID altogether, and create a new one. If we were to get the RAID to work, i'd always be worried of it's failure again. So i'd like to start from scratch.
That said, first we need to try and get that data, so first off what's are backup situation? I assume not good....
1. What i would like to suggest is that you load up diskwarrior and try to mount either of the drives. If one of them does mount do you have an external firewire drive you can copy the contents to, so we can remove the problem of data being lost?
2. Now back at the terminal promt, type - diskutil list , post the results.
3. type diskutil checkraid, post the results.
Terry
I know how stressful dealing with RAID failures can be. Let's see if we can't move this forward some.
First off, my thinking here is that we want to destroy the RAID altogether, and create a new one. If we were to get the RAID to work, i'd always be worried of it's failure again. So i'd like to start from scratch.
That said, first we need to try and get that data, so first off what's are backup situation? I assume not good....
1. What i would like to suggest is that you load up diskwarrior and try to mount either of the drives. If one of them does mount do you have an external firewire drive you can copy the contents to, so we can remove the problem of data being lost?
2. Now back at the terminal promt, type - diskutil list , post the results.
3. type diskutil checkraid, post the results.
Terry
mmontgomery - Aug 22, 2005 - 10:32 am
Diskwarrior requires mounted drives, and since neither drive would mount, Diskwarrior could not do anything. The technical support for Diskwarrior is GREAT!!! The fellow there wanted to help me figure out why my drives would not mount. We finally did a raw dump of the drives. Both drives had been zeroed out for nearly the first gigabyte! Needless to say, we can forget directory recovery.
My only option now is to use my last backup, and maybe do file recovery on the remaining 159 GB to see if there is anything worth salvaging.
My theory is this. The RAID was a v1 formatted on Panther. Tiger can read RAID v1 just fine, even though Tiger creates v2 RAID. One of my disks probably experience corruption, perhaps due to a crash. This is probably what caused the RAID to be marked degraded.
When I ran Diskwarrior, it was able to find the data, no problem, even though I could not get the disk to boot. On further inspection, even though the drives were mirrored, only one drive had a boot segment, which was probably the drive that was spared out due to corruption.
Here is where I made a HUGE mistake. I should have immediately attempted to mount the RAID as a Firewire volume and back up everything before I did anything else while I could still see the data.
Instead, I went ahead to use Disk Utility to repair the mirror. (I think in my original description of events I had this backward; in retrospect, I think I ran disk warrior before I tried to repair the RAID.) Because after I tried to repair the RAID, both drives were marked offline.
I suspect that Tiger has a bug when attempting to repair a V1 mirror, and this resulted in trashing my disks, because my disks were not marked offline until after I attempted to rebuild the mirror. Perhaps Tiger just blindly assumed the disks were V2 raid, and the formats are different enough for Tiger to mess up my raid.
Anyway, I hope my loss serves as a cautionary tale to keep someone else from making the same mistake. And I hope Apple will take this report seriously and fix this bug.
My only option now is to use my last backup, and maybe do file recovery on the remaining 159 GB to see if there is anything worth salvaging.
My theory is this. The RAID was a v1 formatted on Panther. Tiger can read RAID v1 just fine, even though Tiger creates v2 RAID. One of my disks probably experience corruption, perhaps due to a crash. This is probably what caused the RAID to be marked degraded.
When I ran Diskwarrior, it was able to find the data, no problem, even though I could not get the disk to boot. On further inspection, even though the drives were mirrored, only one drive had a boot segment, which was probably the drive that was spared out due to corruption.
Here is where I made a HUGE mistake. I should have immediately attempted to mount the RAID as a Firewire volume and back up everything before I did anything else while I could still see the data.
Instead, I went ahead to use Disk Utility to repair the mirror. (I think in my original description of events I had this backward; in retrospect, I think I ran disk warrior before I tried to repair the RAID.) Because after I tried to repair the RAID, both drives were marked offline.
I suspect that Tiger has a bug when attempting to repair a V1 mirror, and this resulted in trashing my disks, because my disks were not marked offline until after I attempted to rebuild the mirror. Perhaps Tiger just blindly assumed the disks were V2 raid, and the formats are different enough for Tiger to mess up my raid.
Anyway, I hope my loss serves as a cautionary tale to keep someone else from making the same mistake. And I hope Apple will take this report seriously and fix this bug.
MacIT - Aug 22, 2005 - 10:42 am
Michael,
That's an unfortunate end to the story, but i'm glad you were able to find some answers.
Thanks for passing on the conclusion as well.
Terry
IF THIS IS YOUR QUESTION AND YOU WISH TO RESPOND, LOGIN HERE FIRST.
