Logs spammed with "Buffer I/O error on dev nvme0n1p1, logical block"

[Tritech]

I've passed through my entire NVME drive/controller to a VM and the unraid logs are just filled with these lines. I've tried chkdsk in windows with no errors and crystaldiskinfo shows none either. Other than the logs, the VM seems to run fine. Should I be worried, or is there a fix?

Feb 3 22:52:13 Tower ntfs-3g[24390]: ntfs_attr_pread_i: ntfs_pread failed: Input/output error
Feb 3 22:52:13 Tower ntfs-3g[24390]: Failed to read of MFT, mft=83212 count=1 br=-1: Input/output error
Feb 3 22:52:13 Tower ntfs-3g[24390]: Could not decode the type of inode 83212
Feb 3 22:52:13 Tower ntfs-3g[24390]: ntfs_attr_pread_i: ntfs_pread failed: Input/output error
Feb 3 22:52:13 Tower ntfs-3g[24390]: Failed to read index block: Input/output error
Feb 3 22:52:13 Tower kernel: Buffer I/O error on dev nvme0n1p1, logical block 807235, async page read
Feb 3 22:52:13 Tower kernel: Buffer I/O error on dev nvme0n1p1, logical block 12251, async page read

 

[itimpi]

Are you sure that you do not have the drive mounted in Unassigned devices when you start the VM?   The error message suggests that Unraid is trying to use the file system on the disk rather than ignoring it and letting the VM handle it.  You cannot successfully pass through a mounted drive.

[Tritech]

It was my understanding that a mounted drive can be "released" and passed through bare metal, and then it's returned to the unassigned drives list when the VM is shut down. I probably had the drive mounted when the VM was started. I'll test and see if not mounting first makes a difference.