Hey guys,  

 

I've seen this happen a few times while searching for my answer.  It looks like I may need to add a quirk for these NVME drives I just added.  Wondering what the best way to go about doing that is.   If more info is needed just let me know.

 

[   50.604707] nvme nvme0: pci function 0000:04:00.0
[   50.615493] nvme nvme1: pci function 0000:0b:00.0
[   50.616039] nvme nvme2: pci function 0000:13:00.0
[   50.616604] nvme nvme3: pci function 0000:1b:00.0
[   50.686324] nvme nvme1: 8/0/0 default/read/poll queues
[   50.693917] nvme nvme3: 8/0/0 default/read/poll queues
[   50.695422] nvme nvme0: 8/0/0 default/read/poll queues
[   50.697167]  nvme1n1: p1
[   50.697568] nvme nvme3: globally duplicate IDs for nsid 1
[   50.697647] nvme nvme3: VID:DID 1cc1:622a model:ADATA LEGEND 960 firmware:A200V4TM
[   50.700033] nvme nvme2: 8/0/0 default/read/poll queues
[   50.702759]  nvme0n1: p1
[   50.704592]  nvme2n1: p1

 

04:00.0 Non-Volatile memory controller: ADATA Technology Co., Ltd. Device 622a (rev 03)
0b:00.0 Non-Volatile memory controller: ADATA Technology Co., Ltd. Device 622a (rev 03)
13:00.0 Non-Volatile memory controller: ADATA Technology Co., Ltd. Device 622a (rev 03)
1b:00.0 Non-Volatile memory controller: ADATA Technology Co., Ltd. Device 622a (rev 03)

 

It's also worth mentioning that i'm passing these pci devices through ESXi to unraid (individually) so if there is something I can do on that side to address without making to unraid, I'm happy enough to do that.

 

I'm running unraid 6.12.0-rc5.

 

 

Does it act the same booting Unraid directly?

It's actually a different one that doesn't show up when I do that.

 

[   50.648914] nvme nvme0: pci function 0000:04:00.0
[   50.654124] nvme nvme1: pci function 0000:0b:00.0
[   50.654440] nvme nvme2: pci function 0000:13:00.0
[   50.654754] nvme nvme3: pci function 0000:1b:00.0
[   50.741506] nvme nvme2: 8/0/0 default/read/poll queues
[   50.741819] nvme nvme0: 8/0/0 default/read/poll queues
[   50.742721] nvme nvme3: 8/0/0 default/read/poll queues
[   50.745882]  nvme2n1: p1
[   50.746207]  nvme0n1: p1
[   50.746490] nvme nvme1: 8/0/0 default/read/poll queues
[   50.749391] nvme nvme1: globally duplicate IDs for nsid 1
[   50.749472] nvme nvme1: VID:DID 1cc1:622a model:ADATA LEGEND 960 firmware:A200V4TM
[   50.811802] BTRFS: device fsid 63cbca3e-40c2-4e71-83a7-488ca66bebc6 devid 2 transid 61 /dev/nvme2n1p1 scanned by udevd (782)
[   50.886278] BTRFS: device fsid 63cbca3e-40c2-4e71-83a7-488ca66bebc6 devid 1 transid 61 /dev/nvme0n1p1 scanned by udevd (782)

 

this is worse because it breaks my cache pool.

after playing around with this, its just that 1 & 3 are conflicted to whichever one loads first, i get (this could be random).

 

same experience in ESXi vs unraid directly,  in ESXi i've just stopped passing through the drive I don't use (1 in this case) so i always get 3 to work and my cache is fine.  I'll pass it back through later if I can get a fix in there.

Added to the list, quirk should be added for next Unraid release.

Would you please consider the patch below?

 

I have Minisforum UM790 Pro with AMD 7940HS SOC and 2xSSD Fanxiang S880 2TB SSDs.

This patch helped kernel to recognize the second SSD.

 

amd.patch

8 hours ago, distasteful-reputation1044 said:

I have Minisforum UM790 Pro with AMD 7940HS SOC and 2xSSD Fanxiang S880 2TB SSDs.

Please post the diagnostics.