How do I restore a failed drive after improper shutdown?

[thequinox]

My house lost power and the system was improperly shut down. Usually when this happens I just have to do a parity check when starting the array back up. I had a UPS but it died, probably time to get a new one. Anyway, this time upon boot all my drives were unassigned. I did some searching on the forum, ran a checkdisk on my flash drive, it's fine. The .dat file used to house the assigned drives however was 0 bytes.

 

I deleted the file and went to start assigning the drives back, but noticed I was missing one. Looked at the syslog and found that one drive was failing to respond. It detected something was there, but first tried 3Gbps, then 1.5Gbps, then it timed out. So looks like I lost a drive in the power failure as well.

 

Now, I have an extra drive I can put in. I slid it in and it was detected no problem, however because it was an improper shut down my parity isn't trusted. I know the fileserver wasn't doing anything at the time because it was in the middle of the day and no one was home, so I personally think the parity is fine. Do I put in the new drive, and just hit rebuild? I'm worried that if I do that the parity drive may be overwritten with the blank drive instead of the other way around.

 

What are the correct steps to go through to prevent any data loss? Thanks guys.

 

Update: I was able to borrow a 2TB drive, rebuild with it, and then expand with my spare 4TB. The server is now up and running. Many thanks to johnnie.black and RobJ!

[thequinox]

Bump for visibility. Hoping someone knows the correct course of action?

[JorgeB]

Do you have a super.dat backup?

 

Also, post your diagnostics.

[thequinox]

I looked at my last backup, but unfortunately I've added another drive since then, so I don't think it will be of any use. I have attached the requested diagnostic file. I think I just need to tell unraid that the parity is valid, then the rebuild will proceed as if there was a normal disk failure. Just not sure the right way to go about that.

fileserver-diagnostics-20160514-2116.zip

[JorgeB]

You can try this, I think it's your only chance, it should work if you didn't mount any disks after the failure and the replacement disk is the same size of the failed disk.

 

http://lime-technology.com/forum/index.php?topic=47486.msg454776#msg454776

[thequinox]

Thanks for the link Johnnie. I did a search before posting but I didn't come across that one. It's almost exactly what I'm facing so thank you for getting me the link.

 

It's interesting that his super.dat failed as well. Is that a known bug with unRAID 6? I'll have to do some searching. 

 

Lastly, does the new drive have to be the exact same size as the failed disk? I have a brand new 4TB sitting here ready to go in, wouldn't the rebuild procedure be basically the same as the expansion procedure?

UnRAID Manual 6

When you Start the array after replacing a failed disk, the system will reconstruct the contents of the failed disk onto the new disk; and, if the new disk is bigger, expand the file system.

 

Thanks again.

 

[JorgeB]

There's no problem rebuilding a failed disk to a bigger one in a normal situation, that's not your case, you have to do a new config and trust parity with a missing disk, this is not a supported procedure.

 

The procedure I linked should work, and it did in fact work in my tests and for at least 2 users in the forum in a similar situation, I did test a rebuild to a bigger disk and while it also worked, there was some file system corruption, so I wouldn't recommend it.

[thequinox]

Sorry for the late reply. I have procured another 2TB drive and attempted the instructions you linked to, but ran into a problem.

 

The new drive shows as unmountable, as expected, but so do two of my other good drives. The 2TB assigned currently as Disk 5 is the one I am using to replace the failed drive, but I haven't had any problems with mounting the other 3TB drives in the past.

 

If I recall correctly one of the 3TBs was XFS, if that makes any difference. The slots they're assigned to don't matter as long as the correct parity disk is selected, correct?

fileserver-diagnostics-20160520-1714.zip

[RobJ]

It tried to mount them as ReiserFS, which they apparently aren't,  so stop the array and change their file system to the correct one for each.

[thequinox]

In the disk properties, one was set to RiserFS and one was set to XFS, as expected. I switched the RiserFS to XFS, and the XFS to RiserFS, and that did the trick. So somehow they were just mixed up.

 

I'm starting the rebuild to the 2TB now. Thanks a lot guys!