NVMe Support for SMART reporting on VMware vSphere ESXi

[mc1903]

Hi,

I am trying to use smartctl 7.2 2020-04-23 to interrogate my Samsung 970 Pro NVMe PCIe M.2 drives and I am having no luck.

I compiled the executable on an Ubuntu machine, but I am running it on a VMware vSphere ESXi 6.7 host server.

I have two 970 Pro devices installed:

[root@mc-esxi-p-101:/opt/smartmontools] ls -l /dev/disks | grep Samsung
-rw-------    1 root     root      953.9G May 14 21:37 t10.NVMe____Samsung_SSD_970_PRO_1TB_________________40FAB2915A382500
-rw-------    1 root     root      953.9G May 14 21:37 t10.NVMe____Samsung_SSD_970_PRO_1TB_________________40FAB2915A382500:1
-rw-------    1 root     root      953.9G May 14 21:37 t10.NVMe____Samsung_SSD_970_PRO_1TB_________________88FAB2915A382500
-rw-------    1 root     root      953.9G May 14 21:37 t10.NVMe____Samsung_SSD_970_PRO_1TB_________________88FAB2915A382500:1
lrwxrwxrwx    1 root     root          68 May 14 21:37 vml.0100000000343046415f423239315f354133385f323530300053616d73756e -> t10.NVMe____Samsung_SSD_970_PRO_1TB_________________40FAB2915A382500
lrwxrwxrwx    1 root     root          70 May 14 21:37 vml.0100000000343046415f423239315f354133385f323530300053616d73756e:1 -> t10.NVMe____Samsung_SSD_970_PRO_1TB_________________40FAB2915A382500:1
lrwxrwxrwx    1 root     root          68 May 14 21:37 vml.0100000000383846415f423239315f354133385f323530300053616d73756e -> t10.NVMe____Samsung_SSD_970_PRO_1TB_________________88FAB2915A382500
lrwxrwxrwx    1 root     root          70 May 14 21:37 vml.0100000000383846415f423239315f354133385f323530300053616d73756e:1 -> t10.NVMe____Samsung_SSD_970_PRO_1TB_________________88FAB2915A382500:1
[root@mc-esxi-p-101:/opt/smartmontools]

I seem to be able to open the NVMe device successfully:

[root@mc-esxi-p-101:/opt/smartmontools] ./smartctl --device=nvme /dev/disks/t10.NVMe____Samsung_SSD_970_PRO_1TB_________________40FAB2915A382500
smartctl 7.2 2020-04-23 r5053 [x86_64-linux-6.7.0] (local build)
Copyright (C) 2002-20, Bruce Allen, Christian Franke, www.smartmontools.org

NVMe device successfully opened

Use 'smartctl -a' (or '-x') to print SMART (and more) information

[root@mc-esxi-p-101:/opt/smartmontools]

But when I try to show any information about the drive with -x or -a I get the same error:

[root@mc-esxi-p-101:/opt/smartmontools] ./smartctl -a --device=nvme /dev/disks/t10.NVMe____Samsung_SSD_970_PRO_1TB_________________40FAB2915A382500
smartctl 7.2 2020-04-23 r5053 [x86_64-linux-6.7.0] (local build)
Copyright (C) 2002-20, Bruce Allen, Christian Franke, www.smartmontools.org

Read NVMe Identify Controller failed: NVME_IOCTL_ADMIN_CMD: Function not implemented
[root@mc-esxi-p-101:/opt/smartmontools]

I know the 970 Pros support SMART reporting, as I was able to interrogate them in the same machine/hardware configuration when it was previously running a Windows 10 OS, admittedly I was using Crystal Disk Info.

Is there anything I can do to get this working or is the NVMe support for SMART reporting in smartctl just not available yet?

Any help will be greatly appreciated.

Many Thanks
Martin

[Christian Franke]

I compiled the executable on an Ubuntu machine, but I am running it on a VMware vSphere ESXi 6.7 host server. ...

Read NVMe Identify Controller failed: NVME_IOCTL_ADMIN_CMD: Function not implemented

Apparently the ESXi kernel does not implement this NVMe pass-through I/O-control. It is available on regular Linux kernels.

If possible, please ask VMware support if an alternative I/O-control exists. Leaving ticket open as undecided for now.

[Christian Franke]

NVME_IOCTL_ADMIN_CMD is not implemented on ESXi and there is no known alternative.

Feel free to reopen this ticket if information about an alternative pass-through I/O-control is available.