Cannot format NVME cache drive

[toasty_tt]

I recently added a spare NVME drive to my unraid server using a PCIE adapter. It shows up correctly and i'm able to add it to the cache pool, but it says: Unmountable: Unsupported partition layout. I have the option to format, but when i try to do this, the status doesn't change. I checked the disk log and I'm getting some MBR errors:

 

Oct 12 00:05:47 Tower kernel: nvme0n1: p1
Oct 12 00:05:47 Tower root: /dev/nvme0n1: 8 bytes were erased at offset 0x00000200 (gpt): 45 46 49 20 50 41 52 54
Oct 12 00:05:47 Tower root: /dev/nvme0n1: 8 bytes were erased at offset 0xe8e0db5e00 (gpt): 45 46 49 20 50 41 52 54
Oct 12 00:05:47 Tower root: /dev/nvme0n1: 2 bytes were erased at offset 0x000001fe (PMBR): 55 aa
Oct 12 00:05:47 Tower root: /dev/nvme0n1: calling ioctl to re-read partition table: Success
Oct 12 00:05:47 Tower kernel: nvme0n1: p1
Oct 12 00:05:47 Tower  emhttpd: writing MBR on disk (nvme0n1) with partition 1 offset 2048, erased: 0
Oct 12 00:05:47 Tower  emhttpd: error: mkmbr, 2187: Input/output error (5): write: /dev/nvme0n1
Oct 12 00:05:47 Tower kernel: nvme0n1: I/O Cmd(0x1) @ LBA 0, 8 blocks, I/O Error (sct 0x0 / sc 0x4) MORE 
Oct 12 00:05:47 Tower kernel: I/O error, dev nvme0n1, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 1 prio class 0
Oct 12 00:05:47 Tower kernel: Buffer I/O error on dev nvme0n1, logical block 0, lost async page write

 

The drive worked fine when it was in my pc, and the smart report I generated came back with no errors. The drive is still formatted as NTFS with my old data on it (that i've already moved).

[JorgeB]

2 hours ago, toasty_tt said:

Oct 12 00:05:47 Tower  emhttpd: error: mkmbr, 2187: Input/output error (5): write: /dev/nvme0n1

This is a device error, you can try:

first reboot, then wipe it  with

blkdiscard -f /dev/nvme0n1

then try again to format

[KptnKMan]

Oh thanks for this solution, I've been getting similar issue where I replaced/upgraded my NVME cache with a larger drive, and reassigned the existing cache disk to an Unassigned disk. As a result, for some reason the "old" cache disk (now 2nd NVME) would show up with the option to FORMAT, but after that would only present the option to Preclear, which I did, but then I cannot utilise it as an Unassigned disk because the MOUNT option never becomes usable:

 

After formatting it and Preclearing it a few times, nothing seemed to be working.

I also cleared the disk a few times, deleting all partitions.

After clearing the disk a final time and performing a full system reboot, it seemed to become usable again:

 

Seems like a odd bug somewhere, maybe?

[toasty_tt]

Hi,

I've tried the command you've given me, and then rebooting, however, I'm still getting the same MBR errors.

[JorgeB]

Please post the diagnostics after running the command on a fresh boot.

[toasty_tt]

tower-diagnostics-20221013-0039.zip

[JorgeB]

Post output of:

fdisk -l /dev/nvme0n1

 

[toasty_tt]

I decided to take the drive out and make the MBR via disk management, and then put the drive back in the server, and I'm able to use it as a cache drive without issue now.

[TheFiuz]

Same issue here. No matter how may times I delete partitions, clear disks or issue blkdiscard, the existing nfs partition on the first drive won't be replaced by the BTRFS one I create and in the second one, no partition is created altogether (same BTRFS).

 

[JorgeB]

On 10/13/2022 at 8:16 AM, JorgeB said:

Please post the diagnostics after running the command on a fresh boot.

 

[TheFiuz]

I rebooted to run the diagnostics... and I could format the drives. 🤷‍♂️

[JorgeB]

Fresh boot sometimes helps, that's why I ask for that.

[KptnKMan]

On 10/13/2022 at 11:32 AM, toasty_tt said:

I decided to take the drive out and make the MBR via disk management, and then put the drive back in the server, and I'm able to use it as a cache drive without issue now.

I am assuming that you rebooted the server during this operation?

Ie: shut down, remove disk, format in another machine, repace in original machine, startup?

 

Since this is marked as the solution, I wonder if its the reboot that is important?

Also, is there a way to reset this without rebooting?

Edited October 14, 2022 by KptnKMan