I woke up one morning last week and found my unraid server was unresponsive, I pulled the plug out to force a power cycle, and afterwards my Unraid came up but my disk share was gone and also on Docker I see an error telling me the docker services won't start. I then tried to repair the array, upgraded to Unraid 7 (yes dumb move I know) and disk 2 still shows disabled. Finally I did the trick to stop array, unassign disk 2, start the array, stop again and assign disk 2, and the rebuild finally finished today but still nothing on my shares. I know I've done quite a few silly steps and have only myself to blame, but does this mean my data is completely gone or at least there is some steps that I should try to salvage anything. I have a 2 disk array set up with both plugged into a minicomputer via USB 3 connection and disk 2 (the one with JMicron name) itself is a disk array with 4 disks inside. Diagnostic logs attached here.

azunraid-diagnostics-20250212-0955.zip

Check filesystem for disk1

Wow thanks a lot @JorgeB! I am now able to see my files....I think. The SMB mount is still not there and it says under the "SHARE" tab: "There are no mountable array or pool disks - cannot add shares.", after I started the array again in "normal" mode I ended up having a red cross next to disk1 and when I hover my mouse on it, it says "Device is Missing[Disabled], Contents Emulated". However there is a square with an arrow pointing to top right next to the red cross that allows me to browse the contents of /mnt/disk1, and from there I can see the files seems to be there, although when I tried to click on one of the .txt files it says "Preview Not Available".

 

I've run another diagnostic just now and attached here, could you advise me on the next steps I should take? Is it a good idea to run Parity Check with Write Correction? Also I noticed the SMART report in my first parity disk (the WD one) seems to have a few fields that are in "pre-fail" status, is that a good thing (meaning hasn't failed yet) or is it a bad thing? (meaning it is about to fail)

 

Also FWIW I ran the check on disk1 3 times to make sure there is no more error with Disk 1...

azunraid-diagnostics-20250212-2039.zip

Edited February 13, 20251 yr by Anyi Zhu

You only had both parity disks assigned, disk1 was not assigned, so it's being emulated, what happened to disk1, is it still connected?

7 hours ago, JorgeB said:

You only had both parity disks assigned, disk1 was not assigned, so it's being emulated, what happened to disk1, is it still connected?

Mmm....as far as I know there is no other USB storage devices connected to it (other than another tiny USB stick for the Unraid OS itself, I thought disk1 is the XFS partition created by the two parity disks, is it not?

Not sure what you mean, disk1 is being emulated by parity, but that is for when a device fails, you don't start with parity without data devices.

So if I only have the WD 12TB disk and the JMircon 8TB disk attached I should pick the WD Disk as Parity Disk and JMicron as disk1? Assuming there is no other physical disk attached other than the Unraid OS thumb drive.

Correct, or create a pool with both.

Thanks, I will try that and report back. Question - is there any way I can possibly check how the disk configuration was laid out, say, more than 2 weeks ago (knowing that I've rebooted the mini-computer a few times since then)? I basically can't remember how I set out the disk layout, and when I just use one (the WD 12TB disk) as the parity disk and the JMicron one as disk 1, the array won't start (the start button was greyed out), but I also don't think I used WD as disk1 and JMicron as disk2 since I clearly have had some sort of parity set up in place before...so now I just have no idea how I did it before (and my old brain is not very good at remembering anything happened more than 2 days ago let alone more than 2 wks/months since I last played around with the disk config).

The only valid setup is the smaller disk as disk1, and the larger one as parity, the 8TB device is named RAID5? Note that using a raid5 volume with Unraid is a terrible idea.

I have tried several combinations and seems the only way for me to start the array (i.e., the start button is not greyed out) is I use the 12TB WD Easystore disk as parity and JMicron 8TB as parity2, if I leave WD 12TB as parity and JMicron 8TB as disk1 then I ended up getting this error:  Too many wrong and/or missing disks! If I try to set the JMicron as disk1 and WD EasyStore as Disk2 (and leave nothing for Parity and parity2) I ended up getting this error:  You may not add new disk(s) and also remove existing disk(s).

 

The 2TB WD black drive was the internal Windows OS disk in the mini-computer and it has no Linux partitions so I don't think I've ever added it as part of the Unraid setup at all...

 

 

Very confusing how you could have gotten to this state.

 

Did you perhaps start the array without disk1? So disk1 would be emulated by parity.

 

Then later assign that disk as parity2 instead of as disk1?

 

 

On 2/12/2025 at 10:22 AM, Anyi Zhu said:

via USB 3 connection

Possibly this is the root of the confusion. USB not recommended for assigned disks for many reasons.

 

Maybe it didn't recognize disk1 because of USB, so it was missing, then you assigned it as parity2. So both parity are emulating disk1, and emulated/missing disk1 does have 1.7TB of data according to the diagnostics.

 

Do you have another copy of anything important and irreplaceable? You could copy data from the emulated disk1 to somewhere off your "server" if needed.

 

It won't let you change the assignments now without New Config. But, New Config would normally rebuild parity so disk1 would no longer be emulated.

 

We might be able to force it to let you reassign parity2 as disk1 and rebuild disk1, but you would have to follow instructions exactly.

 

Let us know when you are ready to try.

Another possibility would be to rebuild disk1 to a different 8TB disk if you had one.

I think @trurl is right....after hours of tweaking I think I finally have some idea what I have done....I think...Basically I had the 8TB JMicron disk as disk1, and when I rebooted the UnRAID server I guess I probably changed it to parity2 (WD Easystore with 12TB was always as parity disk), and then I rebuild the array...

 

What I am doing now is basically I am trying to start the array with both as parity disk and then unassign JMicron 8TB from parity2 and restart with just the 12TB WD disk, in this particular case it would tell me the array is unprotected but there is a checkbox I can select that would allow me to start the array, and then I stop the array again and add back JMicron as disk1 and restart, now it gave me a warning the contents on the 8TB JMicron drive would be formatted when I started the array but once I do it seems to rebuild the array again.

 

It seems the array is being rebuilt now (will still take another 10 hrs to complete), and at least I am able to see my shares coming back and my SMB are mounted on my Windows client PCs just like before, I am not exactly sure whether this would give me all the data back (I am seeing /appdata, /domains and /isos having a red exclamation mark next to "Cache" under the Storage column, but it could be because the array is rebuilding?), but I think even if I lose those 3 it won't really be the end of the world. I don't run any VMs on this minicomputer (since it is using an Intel N200 CPU and is struggling just as a poor man's NAS) so if I have to guess /domains is empty, I am not sure if there is anything in /isos either, the /appdata one is probably a bit more problematic as my ActualServer (a budgeting app) is running on a Docker container with data directory in /appdata, but that's something I have really only used once if I remembered correctly...

 

I will report back tomorrow when the array rebuild is done, but for now my Samba share looks good, and that's the most important one to me - I haven't opened every single folder to check yet but the one or two folders I have checked so far seems to still have my files. And I think I finally started to understand how the parity and parity2 works...basically if I understand correctly, disk1-24 are like the "primary" storage, parity is like the "secondary backup" and parity2 is like the "tertiary backup", is that correct? I was thinking parity and parity2, when combined, is the "primary" disk, which might have caused so much confusion....

 

RE: RAID5, I believe that's what the box originally came with but I think I have since changed it to RAID10 configuration (because I should have 4x4TB HDDs in there), not sure why it didn't update the disk name though...as for the USB, yes, agreed it is not the best solution but doubt I would be allowed by my better half to spend money build a real RAID tower (especially since I am spending the Valentine's day evening worrying about the UNRAID box :))

1 minute ago, Anyi Zhu said:

it gave me a warning the contents on the 8TB JMicron drive would be formatted

Hopefully it didn't actually say "formatted". It should have said something like "overwritten".

 

If you post new diagnostics we can see whether emulated/rebuilding disk1 still has its data.

Oh you can see it now? Sure here you go! Thanks!

 

And yes, sorry I think overwritten is the correct word, I was typing based on what I remember and my memory, as u can probably tell by now, isn't really the best

azunraid-diagnostics-20250214-2303.zip

Parity is not a backup in any sense, and neither is parity2. This is true even on other systems that implement parity in some way. Parity just gives an extra bit that allows a missing disk to be calculated from all the other disks.

 

https://docs.unraid.net/unraid-os/overview/nas/#parity-protected-array

Looks like emulated disk1 has 1.7TB which is exactly what will be rebuilt.

Nice! Thanks for the doc URL, will take a read!

 

And second part is EVEN BETTER! Thanks so much, I was still a little nervous earlier but now I guess I can finally go to bed and have a nice sleep, hopefully when I wake up the rebuilding is done, and my UNRAID would be just like before! Thanks for all your help!

 

Actually just took a closer read on the doc you shared (thanks again!), it is a pretty simple but genius idea! I guess the UNRAID could really take maximum 1 of the disks to fail, if there is more than one (let's say in the example in the doc both disk2 and disk3 failed) then it wouldn't really be able to recalc the missing bits then, is that correct? Also for parity and parity2, do they work as redundancies to each other (i.e., the information stored on both are identical) or they just keep track of different parts of the "sums" of bits for the "disk" drives?

Edited February 15, 20251 yr by laspammy

Since you only have 1 data disk, parity actually is a backup since it is the only other disk. And if you had parity2 it would also be identical to disk1. That is simply because of the way the calculation works out.

 

But when you have more than 1 data disk, parity is not a backup, and parity2 is not the same as parity. They are just different ways of calculating a missing disk from all the other disks.