DrivePool borked, has disks confused?

[JCMarsh]

My DrivePool was:

1x DC520 12TB

2x WD80EFBX 8TB (0BUK and MIEG)

1x WD80EFZZ 8TB (HDRK)

 

One disk (HDRK) was slower than the others and I wanted to repurpose it in another machine, so last night I removed it from the pool. This should have left me a pool consisting of the two WD80EFBX and the DC520. It didn't turn out that way this morning....

This morning I verified there was no disk activity, and DrivePool showed the old disk (HDRK) was available to add, so I shut down and removed the old disk.

Boot back up and now DrivePool is complaining that a disk (0BUK) that hasn't been removed is missing. Funny thing is, it shows said disk TWICE, but only one instance of it as missing.

Another disk (MIEG) which is part of this same pool, is still installed, but DrivePool doesn't show it at all, and doesn't appear to be complaining that it's missing. It also doesn't show it as available to add to the pool. In Windows the disk is shown as offline due to collision.

Minus what's been backed up in the last week, this pool houses all of my data. Right now, Windows and DrivePool only correctly identify ONE of my three pool disks, with the other two apparently identified as the same disk. How do I proceed?

[Shane]

If they're offline due to collision that suggests that somehow they've ended up with the same "unique" disk ID (a serial number assigned to a drive when initialised by Windows). Normally such collisions only happen when cloning a disk.

( @Christopher (Drashna) I can't imagine DrivePool having any need to change a physical disk's ID - any ideas what could cause this? Windows? )

You can use the Windows Disk Management utility and the DISKPART tool from an Administrator command prompt to view the ID. Find the disk numbers in the former, then use SELECT DISK number followed by UNIQUEID DISK in the latter, to select each disk in the pool in turn and view their IDs. They should be in the form of a long dash-separated hexadecimal string. The pool itself uses an eight character hex string.

If you find two of your disks in the pool have the same ID, you can change one of them via selecting it (as above) and using UNIQUEID DISK ID=newstring command where newstring is a string chosen by you in the same format (e.g. just use the same string but change one of the characters); restart the computer afterwards. Be very careful here that you change only the ID of the particular disk you want.

If you don't find two disks with the same ID then I would recommend opening a support ticket with StableBit.

Regardless, your data should still be intact in the hidden poolpart folders on each disk and could be moved to a new pool if the old one can't be restored.

[JCMarsh]

I did open a support ticket. Drashna suggested the drivepool troubleshooter and reset drivepool to defaults.

I also marked the nvme boot drive dirty and rebooted.

After that, Disk Manager showed all three disks correctly, but DrivePool showed one pool disk as a non-pooled disk.

DrivePool eventually finished duplication this morning, and I started re-enabling services after removing and making a new L2 cache.

It's working now as a 2-disk pool. I'm going to wait until BackBlaze is finished before doing anything else with DrivePool, but may re-add the excluded disk when it's done.

At least I have the pulled WD80EFZZ drives if I need to replicate everything before burning the homeserver down and starting over....

[JCMarsh]

Removed the DC520 today, that went fine. Then it dropped the same disk (M1EG), but it came back after a reboot.

I don't know what the issue is, but apparently DP is giving a two-fer deal on removing 1 drive from the pool. Well, sort of...

It's absolutely amazing that ANYTHING has ever functioned on windows, but I can't continue to use a storage system and OS that are threatening my data. I left windows for ubuntu last year on my desktop and laptop, and the homeserver was my last windows machine. Once my array is up I'm moving everything over to unraid and have a few beers while I DBAN the old windows boot disk...

[Shane]

I'd suggest searching for "unraid" in the forums here, as I've seen posts about people moving in the other direction - unraid to drivepool - after encountering problems with the former. Maybe run them side by side for a while, or at least be prepared for teething issues? In any case, best of luck!

[JCMarsh]

Nah, not worried about unraid eating my data. Bit-rot is a real problem with most storage technologies and systems, not just unraid. There are various ways to guard agains bit-rot. Keeping multiple copies is one way, IF those copies are regularly compared. I appreciated the data availaibility that drivepool provided. I kept everything pool duplicated x2, and never "lost" any files that I know of, but I did still have file corruption happen, ruining several photos (that I know of). That said, it still didn't protect me from bit-rot. I can appreciate the protection that parity provides, as well. It'll be a long game to figure out which one wins, but I'm bettting Microsoft breaks off and falls into the pacific ocean before I lose data due to array trouble......

The new setup is running well. Today I installed a new 12TB red plus, which is settling in as the data disk now paired with the 12TB DC520 as parity disk. Rebuild should be done sometime tonight. After this, I will keep it as a 2-drive, 12TB array behind a 2-disk, 1TB cache pool. I don't qualify as a data hoarder, as I have barely more than 2TB on-hand after years of collecting music and movies, and I don't even use any "arrrs", so this 2-disk setup should serve us well. The vast majority of data on my array is media, so it's replaceable, really. All "can't lose" files are replicated with syncthing between my laptops and desktop, as well as an archive of the same on the array.

Wow, people move from unraid to windows? I can't imagine it. Years ago when I was still using Windows Home Server 2011 I ran a trial of unraid and really liked it. I didn't need it then but figured I'd move to that with my next homeserver build. I never did. When Windows 7 (and server 08r2, whs2011) support ended, I lazily rolled right on over to Windows 10. It worked okay like that, so I let 'er eat until recently, when it seemed the dang thing was starting to implode. Windows' UI was getting slow, DrivePool might open in a minute, syncthing required some manual tending here and there, Plex wasn't the reliable stalwart it had always been... it's health wasn't looking good.

In years past I'd have DBAN'd the boot disk and start fresh with a clean install, but I just couldn't put myself through that again. I'll give unraid a fair shake and if I have to leave unraid for any reason I will definitely be on a GNU/Linux system. If it gets bad I guess I'll have to learn ceph and configure my own storage solution.