SOLVED Saving a Dying Drive on 8.3

[mankyd]

I have a drive that was about 3 years old, hosting a single ZFS partition all by itself. The other day, I noticed that it was misbehaving - FreeNAS was reporting that the status of the drive was "Unknown", and I couldn't get any other useful info from it.

I tried mounting via NFS. Generally speaking, I would get I/O errors. But if I was patient (i.e. I'd run "ls", then wait 10 minutes, then run "ls" again) I could see the data on the drive over the network.

I finally broke down and purchased a bunch of new drives to replace it with via RAID. I powered down the machine (probably my biggest mistake); installed the new hardware; and then powered everything back up.

Now the machine fails to completely boot. I see lots of the following error messages repeated over and over and over: "rescan already queued". If I leave the old drive unplugged, it boots fine, and the new drives show up. As soon as I hot-plug the new drive in, the error messages appear and the machine starts slowing down (though it is somewhat responsive). I should add that the new drives are plugged into a new, separate, SATA controller. The malfunctioning drive is plugged into the motherboard.

How can I recover what's on the old drive. I'm running FreeNAS 8.3.

 

[jgreco]

How can I recover what's on the old drive. I'm running FreeNAS 8.3.

You need to get the drive into a state where it is readable by the OS. Alternatively, a data recovery service may be able to recover it onto a replacement drive.

 

[mankyd]

You need to get the drive into a state where it is readable by the OS.

That much I figured out ;). I guess my question is, how do I get it there? It was readable this morning.

 

[cyberjock]

That much I figured out ;). I guess my question is, how do I get it there? It was readable this morning.

Then get in a time machine and go back to this morning. ;)

Ok, I'm kidding. But as your hard drive is in a case of constant deterioration as it is used, I'd expect things to generally only get worse. Unfortunately you've appeared to wait too long and recovery is going to be very difficult, if even possible.

 

[mankyd]

OK, an update. I did a fresh install of 9.3 and moved the SATA connection to the second header on the motherboard. The machine boots now, yay! However, I can't seem to import the volume. When I click "import" it starts the import, works for ~10 minutes, and then the modal disappears.

It does seem to recognize that there is a volume to import. When I click import, I can see it listed.

Anyone know how I can get this old volume imported?

 

[cyberjock]

If it won't import from the WebGUI then ZFS is probably too corrupted to be mountable. :(

 

[mankyd]

Fixed!

First, I connected to it another machine, thinking that the SATA controller on the original machine may be dying. Then, I ran "zpool import" and saw the pool in question listed. Finally, "zpool import -fF <pool_name>". A few minutes later, I had the partition mounted and am currently copying off the data that I can.