Hey guys. first time caller, long time listener. I'm at my wit's end... I'm not a pro at server setup, not by a long-shot. Any help at all would be hugely appreciated.

 

Here's my story (diagnostics attached), please stay with me if you can:

 

My array for quite a while has consisted of 13 drives (parity and 1 thru 12. No cache, I guess I'm patient). About a year ago I'd noticed that my "Disk 10" decided to take on an Egyptian theme and adopted a pyramid icon next to it (running 6.0.2 beta at the time, I believe). All of the contents were still accessible and usable without a hiccup, even though the temperature and some other stats were not reported and the web interface didn't think it was "spun-up". I'd chalked it up as some sort of glitch because it "worked" as far as I could tell and ignorance ensued.

 

Fast-forward to a week ago:  I upgraded to 6.1.9 and Disk 10 then showed a big, red X that, when hovered over, reported "device is disabled, contents emulated". That added to the annoyance that more and more often I would get errors when writing files to the server. I have been stockpiling hard drives so I figured it was time to retire/replace Disk 10 as the physical drive must've taken a dump. That's just what I did... I copied all of the drive's contents to a big ol' external drive beforehand (hooray?), I shut her down, swapped out the drive and fired it back up. I defined the new HDD in Disk 10's spot and ran the super-long clear and then ran the super-super-long data rebuild. Good news: I got my green dot back! Bad news: about 90% of the files were not rebuilt to the drive (though all the folders were still there) and most all of the files that were still present were corrupt in that they would not open at all. These corrupt files also refused to be deleted, overwritten, renamed, etc. FYI: the web interface also thought the rebuilt drive was back up to the original content size (used/free).

 

Again, luckily, I have all of the contents of that a-hole Disk 10 currently backed-up on an external drive so all is not lost. I'd hoped that I could just reformat the drive so I redefined to xfs, rather than the original rfs because that's the only way I could figure out to initiate a reformat... I ended up with an empty Disk 10 with a happy green dot again. Not so thrilled to be "manually" rewriting 4TBs of data to a drive, but I'd accepted my defeat. Then, after just transferring approximately 5GB of data that damned red X came back again! It's back to "emulating" what is basically zero contents.

 

If you're still with me (thank you) my questions are pretty straightforward and short:

 

1) How do I get my parity drive to wipe its "memory" of the sh*tshow that is Disk 10 so it doesn't try to rebuild a worthless piece of garbage on a brand new, perfectly functional and "cleared" drive that I hope/plan to just manually repopulate?

 

2) Is there some other way I have/need to do this? Like I give this new physical drive a different number, say... disk 13, then once populated with the data I want on it, redefine it to become the new Disk 10 somehow? (Reason I want Disk 10 specifically to return is that it contains media that's already defined within a library).

 

Hey... you made it this far! Thank you so much! 

server-diagnostics-20160907-1904.zip

server-diagnostics-20160907-1904.zip

That replacement disk doesn't look that good

Sep  7 18:33:59 Server kernel: md: import disk10: [8,224] (sdo) HGST_HDS724040ALE640_PK2301PBJD7HKT size: 3907018532
Sep  7 18:33:59 Server kernel: md: disk10 replaced

Serial Number:    PK2301PBJD7HKT
  5 Reallocated_Sector_Ct   0x0033   098   098   005    Pre-fail  Always       -       204
196 Reallocated_Event_Count 0x0032   082   082   000    Old_age   Always       -       488

Did you preclear it? That should have told you about its problems.

 

Do you have Notifications setup? That would have told you about its problems also. Notifications are one of the best new features of V6 and can help you catch problems before they become worse.

 

The only way to get parity to forget about the disk is to rebuild parity without it. Then you would have to clear the disk again before you could add it back. So probably easier and quicker to just let it rebuild an empty disk. I wouldn't rebuild to that disk though, or in fact use it in my array for anything. Do you have the original disk? Maybe it is actually good enough to use.

 

And you really should have asked about this a year ago rather than just letting things deteriorate. I wonder if you don't still have some issues with the files you copied off, since it sounds like you let things go till you had filesystem corruption.

Hey trurl. Thank you so much for the quick reply!

 

I didn't actually "preclear" it, as far as I know what that is (preclear requires a process performed using a software outside of a tool or plugin provided in unRAID itself, right?)... I let it run the full-on looooooong clear process that unRAID does and it found this brand new drive to be healthy and problem free. This is a fairly decent brand drive and I just ripped it out of the packaging. HGST Deskstar 4TB 7200RPM Sata III 6Gbps 64MB Cache (not posturing or gloating by any means as I know this isn't a Ferrari, but it's a fair "middle of the road" drive, right?... my parity is the same or similar).

 

I do not have notifications setup. I didn't even know that was a thing. I surely will now.

 

So, if I completely drop/remove the disk from the array (should I power down,  physically unplug the stupid thing and fire back up?) then run a "parity check" does that rebuild parity without it and effectively lose the "old" Disk 10 info or is there a different process I need to find/run?

 

Yeah, I appreciate what you're saying, really I do. I should have come to the doctor the second I felt the tingle in my throat. I conceded that I'm an idiot. All the other drives/data appear to be intact. I'm just thankful that you're here to help this idiot through his idiocy. You're awesome!

 

 

Crap... I forgot to answer a question: I do still have the original disk 10. She's pouting right here beside me.

 

unRAID will not even let you do a parity check with a missing or disabled disk.

 

The part about letting unRAID clear the disk instead of preclearing it doesn't really make sense for a disk rebuild. A replacement disk doesn't need to be clear since it will be completely overwritten anyway. unRAID only requires a clear disk when adding it to a new slot. A clear disk has no effect on parity, so when you add a disk to a new slot it has to be clear so parity will remain valid. But when replacing a failed disk to rebuild, parity will be used with all the other disks to calculate all of the replacement disk's data, and parity will still be valid when it finishes the rebuild since the rebuilt disk agrees with parity. So long story short, if you just assigned the disk to the same slot as the disk you were replacing, unRAID would not clear it.

 

You should use a different disk whatever you do and test that different disk or at least look at its SMART before you try to use it. If you plug the original disk in without assigning it you should be able to get a SMART for it.

 

As I said before, it's probably simpler and quicker to just let it rebuild an empty disk.

 

If you insist on removing the drive instead, and then adding a new drive (or the old original one), here is what you would have to do.

 

1) Set a New Config and reassign all your other drives. It is very important that you do not accidentally assign one of your data disks to the parity slot or it will get overwritten with parity and that disk's data will be lost.

 

2) Start the array and let parity rebuild.

 

3) Stop the array, add the precleared new disk, then start the array and let unRAID format it. Or if you don't preclear it unRAID will take your array offline while it clears it, and then format it.

 

So, as you can see, more complicated and more time-consuming than just letting it rebuild an empty disk.

 

It probably seems like a waste of time to rebuild an empty disk, but the thing you have to remember is an empty disk is not a clear disk. The word "format" means "write an empty filesystem to this disk". That is what it has always meant on every operating system you have ever used. An empty filesystem actually contains data, the metadata of the filesystem which represents an empty top level folder, ready to have folders and files added to it. When you format a disk, unRAID handles the write of the empty filesystem just like it does any other write, by updating parity. So the parity disk actually agrees you currently have an empty disk, but not a clear disk. Removing the empty disk will invalidate parity, so that is why parity would have to be rebuilt. And then a disk that got added after that would have to be clear so parity would remain valid.

 

All this makes a lot more sense (perhaps even obvious) if you actually understand how parity works. It is in the wiki, and it isn't very difficult.

Thank you so very much for your thorough and succinct explanation of unRAID and its workings. What you have explained to me will surely prove invaluable. I'm sure than I've just stumbled and you've set me on the correct path. I appreciate your time and attention and I'll hopefully be smart enough to apply what you've taught me.

 

Thank you!