Marvell RAID 1 crisis

[GnatGoSplat]

I'm having a little crisis with a Marvell RAID 1 setup.
I accessed my home server running Windows Server 2008 R2 just fine all day as I do every day. Then tonight I access it, and WTH, a lot of important files are gone! Just gone! Practically gave me a heart-attack. So I calm myself down, look at Event Manager, and it looks like the computer rebooted itself somehow and thinks it hasn't been run in 2-months. I checked the Marvell utility, and it shows both mirrored drives are Degraded with blinking exclamation marks on them. It looks like the computer has completely reverted back to 2-months ago. I theorize that the only way this is possible is one of the drives has been degraded for 2-months (email notification apparently not functioning), and then somehow a spontaneous reboot caused by who-knows-what has decided the degraded drive with 2-months old data is the valid one.

In the hopes of recovering the last 2-months of data, I've unplugged both drives from the server and plugged them into another machine. Both show up with no partitions and inaccessible. Running EaseUs on one, and it's picking up files that are less than 2-months old so I'm cautiously optimistic, though a full scan will take 5 more hours.

Is there any easy way to access/recover the data on these drives, or is EaseUs the only way? Is there any possible way to restore the machine the way it was?

I'm really not liking that Marvell RAID setup uses a proprietary format on the mirrored drives.

 

[Ketchup]

How old is your backup? I really don't like RAID in a home setup (assuming that's what this is). The main reasons for RAID usually aren't there for home use. I hope EaseUs gets most/all of your data back. It has done well for me over the years. If you do run into ploblems, try an older version before throwing in the towell. Twice within the past year I had tried the latest version and been disappointed.

 

[Anteaus]

All RAID is proprietary. The only success I've heard of with transplanting hardware based arrays is when moving to an identical controller. Software arrays are more forgiving but you still need to replicate the system environment.

http://www.storage-switzerland.com/Articles/Entries/2013/5/13_Demystifying_RAID.html

I'm going to parrot Ketchup. The easiest way to fix this would have been to rebuild the array and restore from backup. A simple nightly sync to an external drive would also have have been useful if your backup window is wide.

The general saying is that important data needs to be in three places or it doesn't exist; the mirror drive of a RAID 1 array doesn't count as a backup.

 

[PliotronX]

You could give tesdisk a shot, it's a command line recovery tool for Windows. What is the smart status showing on each individual drive? They unfortunately sound both in bad shape and sometimes they'll corrupt the partition table so it's not necessarily proprietary raid as all controllers will only accept foreign config from like-controllers. See what Ease us turns up and if you don't get what you need, try that testdisk. Also, are you connecting these directly to a Sata controller in another system or with a USB adapter? If the latter, recovery options are rarely successful over USB just because the USB bus starts flipping out when the faulty disk encounters a bunch of errors.

 

[GnatGoSplat]

Unfortunately, my backup is worse because it's 3-months old.
EaseUs was able to recover files that I needed immediately, but I had to abort the process since I was tying up my wife's desktop with it that she needs for work today. Right now I have one of the drives plugged into my server machine's built-in SATA ports (bypassing the Marvell RAID card) and running SeaTools on it. Short DST passed. Started the long test last night, but woke up this morning to find SeaTools had crashed to a DOS prompt overlaying the SeaTools screen. Possibly a bad sign. I've restarted it and will see if it keeps happening or not.

From my Googling, I get the impression that it's normal for Marvell RAID 1 drives to be a proprietary format that doesn't show up with a normal partition table on other SATA controllers.

I'm spoiled by the AMD Fake RAID I was using before. In that, I could attach either mirrored drive to another PC as an external drive and it was fully recognizable. I had to give that up because I upgraded to a motherboard without RAID capability.

 

[GnatGoSplat]

So I ran SeaTools for Windows on both drives that I pulled from the server since SeaTools DOS kept crashing to DOS prompt. They both fail Long Generic test. One after several hours. The other after just a few minutes. PliotronX, your suspicion both drives are failed seems to be correct.

I hypothesize the computer reverted to 2-months ago, because 1 drive probably failed 2-months ago and the computer was working with the remaining good drive which just failed last night. So with both drives degraded, it couldn't decide which drive to boot off of, and decided to boot off the one that was degraded 2-months ago. Apparently, I didn't have the email notification or warnings configured correctly.

I'm actually quite surprised both drives failed so soon after one another. Seems like it must be a design flaw. They are both Seagate 2TB ST2000DL001, purchased at the same time and both just one month shy of 3-years old.

 

[VirtualLarry]

PliotronX, your suspicion both drives are failed seems to be correct.

I'm actually quite surprised both drives failed so soon after one another. Seems like it must be a design flaw. They are both Seagate 2TB ST2000DL001, purchased at the same time and both just one month shy of 3-years old.

Now you're scaring me, just a little. One of my NAS units has a pair of Seagate retail-boxed 1TB HDDs, probably two, maybe three, years running in the NAS. AFAIK, they are spun up all the time. They are in a RAID-1 mirror.

I do have some backups, but not of the most recent stuff on the NAS.

 

[Ketchup]

... I'm actually quite surprised both drives failed so soon after one another. Seems like it must be a design flaw. They are both Seagate 2TB ST2000DL001, purchased at the same time and both just one month shy of 3-years old.

See, this shouldn't surprise you that much, since they were bought at the same time. I remember taking a computer class in the 90s, and our teacher discussing not buying drives at the same time for reasons such as this. Frankly, drives were lasting longer back then than they are now.

 

[PliotronX]

So I ran SeaTools for Windows on both drives that I pulled from the server since SeaTools DOS kept crashing to DOS prompt. They both fail Long Generic test. One after several hours. The other after just a few minutes. PliotronX, your suspicion both drives are failed seems to be correct.

I hypothesize the computer reverted to 2-months ago, because 1 drive probably failed 2-months ago and the computer was working with the remaining good drive which just failed last night. So with both drives degraded, it couldn't decide which drive to boot off of, and decided to boot off the one that was degraded 2-months ago. Apparently, I didn't have the email notification or warnings configured correctly.

I'm actually quite surprised both drives failed so soon after one another. Seems like it must be a design flaw. They are both Seagate 2TB ST2000DL001, purchased at the same time and both just one month shy of 3-years old.

I've actually seen this happen before, sorry man. Does the freshest drive do anything strange? Noises, causes the PC to lock up when plugged in (even booted off another drive?). EaseUS might be all that you can do, perhaps give OnTrack Recovery a try (both their software and data recovery service) if the data in those two months is valuable enough.

Good to know that Seagate is precise in shoddy manufacturing.

 

[GnatGoSplat]

Now you're scaring me, just a little. One of my NAS units has a pair of Seagate retail-boxed 1TB HDDs, probably two, maybe three, years running in the NAS. AFAIK, they are spun up all the time. They are in a RAID-1 mirror.

I do have some backups, but not of the most recent stuff on the NAS.

Make sure to test the alert system and check them regularly!

I relied too much on the alert system and evidently didn't have it configured correctly. Assuming my hypothesis is correct that the first drive degraded 2-months ago, that would have been plenty of notice had I been aware of it. I should have checked status regularly as well, though on a home system it's easy to forget.

Yeah, it definitely seems like common mode failure is a very legitimate concern in hard drives. Also interesting, is when I first upgraded motherboards and didn't have a RAID card yet, I ran a single drive for close to around 2-months. So it's very possible both drives failed with close to the same amount of hours on them. I may have to see if I can pull that out of the SMART data just out of curiosity.

One drive (the one that SeaTools doesn't show an error on for several hours) doesn't lock up the PC. I think EaseUs will work on it. Deep Scan completed and it appears all my files are available for recovery. The other drive does lock up the PC and sometimes SeaTools can't read its model or serial #, it's the one that failed Long Generic test in just a few minutes.