Drive on the way out?

[CasanovaFly]

I was recently reorganizing some files on my server (transferring them all over the place and between the different disks) and I ran into a bit of a problem.

 

I had disabled my parity drive for faster transfer speeds, and about 75% of the way through everything I woke up to an "I/O error" with one of the disks. Now, it's one of the oldest disks in my array, a 1.5TB WD Green drive that's currently 38149 hours (4y, 4m, 6d, 13h) old. The disk became unresponsive, and when I stopped the array it disappeared. I shut down, jiggled some power cables, said a prayer, and booted back up. It was back! Started the array, started copying again, all good. Then it happened again. Rebooted, it reappeared. I decided at this point to try and re-enable parity and run a parity check and cover my arse, and it did so without issue. So now I have parity and the wonky drive currently reads fine.

 

So my question is: what do you think is going on? A quick SMART test provided no errors. Running an extended SMART test now. Am I looking at a mechanical failure, here? It's about time to add another disk to the array, I just want to know what I should do with this one.

[ashman70]

I don't understand how you could of rebuilt parity when the first thing you say is "I had disabled my parity drive"

 

If your parity drive was disabled, whatever parity you had would not have been current since you had disabled the parity drive. Or am I missing something?

[CasanovaFly]

Poor explanation on my part. I'll change the above to be more clear.

 

Had disabled the parity to take advantage of faster write speeds. Then the drive konked out (which was not the parity drive). Then I re-enabled the parity drive and rebuilt the parity drive -- or, rather, ran a parity-check may be more apropos.

[trurl]

What version unRAID?

[CasanovaFly]

6.2.4

[trurl]

Tools - Diagnostics. Post completed zip

[CasanovaFly]

Extended SMART test: no errors

 

Diagnostics zip attached.

thevelourserver-diagnostics-20170118-1838.zip

[trurl]

... It's about time to add another disk to the array, I just want to know what I should do with this one.

SMART attributes for all drives looks OK. You do seem to have some rather full disks. Since you have a 6TB parity, maybe instead of adding drives you could start replacing some of the older, smaller disks like this one with larger disks.

[Frank1940]

Comment and a quick piece of information.

 

First, the information.  IF you go to 'Settings'  >>>  'Disk Settings' and change the "Tunable (md_write_method):" setting to 'reconstruct write', you will probably be writing as fast with full parity protection as you are having the parity disabled!

 

Second, the comment.  I have had a couple of disks over the years become slow to physically spin-up the patters and this causes the disk to disappear or become unavailable.  (In fact, I have one in my Media Sever right now that the BIOS fails to detect if I enable the BIOS 'Fast Boot' feature.  It actually happened with two different motherboards causing the first one to be returned! About twenty years ago, I had a similar problem in a Windows 3.1 computer.  With that Computer, I had to turn-on the POST Memory check to allow the disk to come to speed before the BIOS tried to access it.) 

[CasanovaFly]

SMART attributes for all drives looks OK. You do seem to have some rather full disks. Since you have a 6TB parity, maybe instead of adding drives you could start replacing some of the older, smaller disks like this one with larger disks.

 

That was the original plan. When I started, I started with what I had (a bunch of Greens) plus three new 4TB Reds. Since then I've added more Reds as I've needed. I'm now at a point where the majority of my stuff is on the server and it's not growing faster than I can add disks, so I'll start looking to replace, especially with this recent snafu. I ordered a new 6TB Black drive (got a great deal) that I'll make my new parity drive, then see where I end up space-wise. Thanks!

 

First, the information.  IF you go to 'Settings'  >>>  'Disk Settings' and change the "Tunable (md_write_method):" setting to 'reconstruct write', you will probably be writing as fast with full parity protection as you are having the parity disabled!

 

An interesting tip, I'll give it a shot. Thanks!

[CasanovaFly]

Added my new 6TB as the parity drive, stuck the old parity in with the rest. Loaded the GUI today (I've been transferring stuff all day, moving stuff around) and found that the problem disk now has a red 'X' with the comment "Deviced is disabled, contents emulated." What am I looking at here?

[Frank1940]

Basically, it means there was a write failure to that disk.  Post up another diagnostics file. And let;s see if someone can figure what the issue is.

[CasanovaFly]

I haven't attempted to write anything to the disk since the last time it was "green," so a write failure (if it's what I think it is) would be odd.

 

I think this is all she wrote for this drive, considering the headaches it's given me recently.

thevelourserver-diagnostics-20170202-0325.zip

[trurl]

Drive isn't reporting SMART so it has lost connection. Check connections and try again with another diagnostic.

[CasanovaFly]

This is what it was doing earlier. I think it's mechanical failure. I'm currently trying to pull the files off the disk (and doing so successfully, at least to start). I'll let the transfer run overnight and report back tomorrow.

[trurl]

This is what it was doing earlier. I think it's mechanical failure. I'm currently trying to pull the files off the disk (and doing so successfully, at least to start). I'll let the transfer run overnight and report back tomorrow.

Since the disk is disabled, unRAID isn't actually reading it. Instead it is calculating its data from parity plus all other disks.

[CasanovaFly]

I figured that was the case, what with the "contents emulated" message. Regardless, I was able to pull all the data off the drive.

 

I'm going to proceed an just pull it out of the array. Will I need to rebuild parity if I remove it permanently?

[JorgeB]

I figured that was the case, what with the "contents emulated" message. Regardless, I was able to pull all the data off the drive.

 

I'm going to proceed an just pull it out of the array. Will I need to rebuild parity if I remove it permanently?

 

You'll need to do a new config, see here:

 

https://lime-technology.com/wiki/index.php/Shrink_array#The_.22Remove_Drives_Then_Rebuild_Parity.22_Method

[CasanovaFly]

Just recently did a new config. Do I just need to do another and indicate parity is still valid?

[JorgeB]

If you're going to removed a drive parity won't be valid anymore.

[CasanovaFly]

So if I remove the drive, yes, I will need to rebuild parity?

[JorgeB]

Yes, follow the procedure:

 

https://lime-technology.com/wiki/index.php/Shrink_array#The_.22Remove_Drives_Then_Rebuild_Parity.22_Method

[CasanovaFly]

Followed the second method. Bypassed the clearing procedure since I already removed what few files remained on the drive overnight. Did a new config, indicated parity was valid, all seems good. Will parity-check once I'm finished moving things around.

[JorgeB]

Followed the second method. Bypassed the clearing procedure since I already removed what few files remained on the drive overnight. Did a new config, indicated parity was valid, all seems good. Will parity-check once I'm finished moving things around.

 

Clearing and deleting are completely different things, your parity is invalid and any drive rebuilt before it's synced will be useless.

[gubbgnutten]

Parity does not sound valid to me...

 

A correcting parity check will be quite slow due to all corrections.