When I stopped the array to reboot the server it showed a message that there was a missing pool disk, and now the disk isn't showing as assigned. 

 

This is the first time I have had these kinds of problems with adding or upgrading a pool disk, still not sure what is going on. 

 

Thank you for all your help JorgeB.

milkyway-diagnostics-20250516-1312.zip

10 hours ago, Rhuarc said:

and now the disk isn't showing as assigned.

That's expected since it dropped offline, you will need to do the same procedure again to reimport the pool with one device, then try again creating the mirror, but with a different NVMe device

Ok, I am once again at a loss.  I basically dropped down to a single device pool again, shutdown, installed the other new drive, then started up, removed the pool, and recreated the pool with the original drive and the new 1TB drive.  Now the pool isn't showing the new drive at all, not even showing it as missing!  Diagnostics included from after the reboot.

 

Would this be easier for me to just remove the cache pool by setting any shares that use it to No and have it move all teh files off, then delete the pool and create a new pool with the 2 new 1TB drives?

milkyway-diagnostics-20250517-0245 (1).zip

Edited May 17, 20251 yr by Rhuarc

Pool is currently a single device, is this the new NVMe device?

 

Model Number:                       SPCC M.2 PCIe SSD
Serial Number:                      20250214A0750

 

Assuming that is a 1TB NVME yes that is the new device.  I thought I added it to the cache pool, but when I started the array it did not add.

Stop the array, change pool slots to 2, assign that device, start the array and post new diags.

Ok, I did that and am continuing to get tons of errors.

 

The PCIE adapter I am using I know is good because the pool was working great before I tried to start this upgrade.  I think I may try just creating a new pool with just the 1 drive and see if I get errors.  I may even then just change shares that are using the old pool to use the new pool, then if that works I'll add the second 1TB drive to the new pool.

 

I have no idea why this would make a difference, but I am at a loss right now.

milkyway-diagnostics-20250517-1546.zip

I am attempting to just get back to my original cache to make sure the motherboard is ok, and now neither one of my original cache drives is showing that they have a file system.  I have included 2 diagnostics, each one with the different drive added as cache.  THe drive ending in 1207 should never have been overwritten.  Is there a way do a repair on the cache drive to get back the data?

 

I think I may have gotten a couple of bad 1TB drives, I am going to order new ones but in the mean time I just need to get back to where my previous cache is accessible.

milkyway-diagnostics-20250518-0201.zip milkyway-diagnostics-20250518-0206.zip

Well, temporarily I was able to mount the cache drive and am currently copying over all of the data that was on it to the array.  Hmm,  Could I point the shares that were pointing to the cache pool change and point them to the array now and have them find the data that I copied over?  Assuming I put it in the correct place on the array of course.

 

Hopefully once I get working 1TB drives copying all of the data back over to a cache pool will go easily enough.

The NVMe device is still dropping, suggesting either there's a problem with it or the board.

 

If the shares are using /mnt/user they will continue to work with the data in the array, if they are using /mnt/cache you need to adjust.

 

 

Well, I have 2 new drives on their way, hopefully I just got a bad batch.  I am also switching from using a PCIE riser to using both onboard m.2 slots as soon as I get the SAS extender so that I can stop using the onboard SATA ports that share lanes with the 2nd M.2 slot.  Hopefully this will solve most of the problems and I will just return the other 2 drives that seem to be causing me problems.

 

On a related note, the SAS expander that I got can be powered by molex, but still needs to be mounted somewhere.  I have kapton tape to put over all of the contacts so that hopefully i can use an x1 slot soley for mounting the card,and the motherboard won't see the card at all.  If the motherboard sees the card as installed then it disables the M.2 slot that I need for the NVME drive. 

 

Has anyone tried this and can verify that it works?

Ok, I finally figured out what the problem was.  I twas the damn PCIe to NVME adapter.  It had been working perfectly fine before (at least I thought it was) and then when I went to upgrade the drive it started throwing all of those errors.  I got a new adapter and just added a new 2TB drive (since I thought the 1TB drives were the problem I got 2 new 2TB drives).  I added it, it did the btrfs balance without any errors.

 

So the moral of the story is, check everything in the chain, even if it was just working!

Could someone confirm for me that my cache is in a state that I can now add a second drive on? I am still having trouble getting the raid1 mirroring to seem to work correctly, so I am going to need to take it step by step posting Diagnostics between each step I think.

Thank you!

milkyway-diagnostics-20250520-1712.zip

It should.

Alright, I am still struggling to get my pool upgraded. I think I have it in a stable tate with 1 480GB drive and 1 2TB drive. THey are supposed to be in raid 1, but when I look at the cache settigns page it says Allocation profile is single. Can someone take a look at my diagnostics and see what is going on?

Thank you!

milkyway-diagnostics-20250530-0013.zip

If you mean the cache pool, it's only using one device currently, the other one is missing:

Pool: cache

Data Metadata System

Id Path single single single Unallocated Total Slack

-- -------------- --------- --------- -------- ----------- --------- -----

1 /dev/nvme0n1p1 356.00GiB 1.00GiB 32.00MiB 90.10GiB 447.13GiB -

2 missing - - - 1.86TiB 1.86TiB -

I am at a complete loss to figure out what is going on. The drive shows up in the bios, it shows up in unraid as online, it shows in the system devices section of unraid.

Any help on steps to take to get this back up and running in raid 1 would be very much appreciated.

Thank you!

Device was already missing at boot, would need the diags from when it was added or dropped.

I rebooted, and during the reboot I went into the BIOS and disabled one feature that had to do with PCIE link power adjustment. I have no idea if that had anything to do with it or not.

Included new diags from right after a reboot.

I guess my next steps will eb to try another new drive, and then finally to replace hte motherboard.

milkyway-diagnostics-20250530-0322.zip

Edited May 30, 20251 yr by Rhuarc

You first need to reimport the pool without the missing device:

on main click on the first device for that pool and then "remove pool"

back on main, create a new pool with the same name and 1 slot

assign the pool device, leave the filesystem set to auto

start the array to import the pool

Then stop array, change slots to 2, add the new device, start array and post new diags.

Ok, I think I may be in the same problem I had before.

7 hours ago, JorgeB said:

You first need to reimport the pool without the missing device:

on main click on the first device for that pool and then "remove pool"

back on main, create a new pool with the same name and 1 slot

assign the pool device, leave the filesystem set to auto

start the array to import the pool

Then stop array, change slots to 2, add the new device, start array and post new diags.

I did this, and the array started saying there was a disk missing on cache2. I followed the instructions exactly, only creating the cache with 1 slot and importing the original drive not the new one. I have included diagnostics from after that start.

I then stopped the array, and the cache already showed 2 slots. I tried to assign the new drive in the 2nd slot and start the array. It says Wrong Pool State cache - too many wrong or missing devices.

milkyway-diagnostics-20250530-1141.zip

On 5/15/2025 at 2:40 PM, JorgeB said:

btrfs dev remove missing /mnt/cache

Forgot to mention, you need to remove the missing deice first, same as before.