PSOD when writing to NVMe datastore

[davjones]

Hello,

I receive a PSOD when trying to upload a VM over 25GB to a datastore on a Micron_7400_MTFD NVMe U.3 1920 GB drive. I've tried uploading the VMDK thru datastore browser and thru an OVF import. After the PSOD, and rebooting the ESXi host, the file size (at crash) is roughly 25GB. The original VM is over twice that size. I can deploy smaller VMs and run them fine to that same datastore. This is a new server. The NVMe drive is on the HCL I'm running ESXi 8.0.2 (22380479 ). I'd like to join the VMUG and deploy vCenter but I want to be sure I can run VMware on this server before I commit.

PSOD message:
VMware ESXi 8.0.2 [Releasebuild-22380479 x86_64]
LINT1/NMI (motherboard nonmaskable interrupt), vmkernel has diagnosed a fatal error.
This may be a hardware problem: please contact your hardware vendor.

Module(s) involved in panic: [vmkernel Version Releasebuild-22380479 x86_64]
cr0=0x8001003d cr2=0x43000008a2000 cr3=0x603000 cr4=0x14216c

System: SYS-E301-9D-8CN8TP
CPU: Intel Xeon D-2146NT @ 2.30GHz
Drive: Micron - 1.92TB Micron 7400 Pro NVMe U.3 SSD
Mboard: X11SDV-8C-TP8F

HCL Info:
SYS-E300-9D-8CN8TP (same mainboard as SYS-E301-9D-8CN8TP)
https://www.vmware.com/resources/compatibility/detail.php?deviceCategory=server&productid=50384&vcl=...

Micron_7400_MTFD NVMe U.3 1920 GB
https://www.vmware.com/resources/compatibility/detail.php?deviceCategory=ssd&productid=50600&deviceC...

A PSOD screenshot is attached.

Can someone point me in the right direction?
Thanks - David

[davjones · Accepted solution]

Update:

I added an ISCI datastore on a synology NAS and was able to import the OVF there without issue. The VM ran fine.

Next I tried using WinSCP to copy the VM files from Workstation to the NVMe datastore without crashing. The VM HW was incompatible so I couldn't test running it.

Then I cloned it in Workstation, changing the compatibility mode to ESXi and successfully uploaded it using Workstation to the NVMe datastore. The VM boots fine and I'm putting it through tests now. I'm running a complex EVE-NG lab pushing the CPU to 80%. Not much load on the filesystem. But so far all looks well.

Seems to be an issue transferring large files to NVMe storage using the ESXi Host Client interface.

Thanks,
David

[maksym007]

Thx for letting us know. I was thinking that probably not compatible NVMe

[davjones]

Apparently the NVMe drive is on the HCL. I don't know why ESXi kept experiencing a PSOD when trying to deploy that Ubuntu VM from an OVF using the ESXi Host Client.

I did deploy the VMware-Skyline-HealthDiagnostics-Appliance and it didn't point out any issues with the drives. It didn't crash either when deploying to NVMe. It was a smaller file size, less than 2GB.

As another test, I downloaded a Microsoft Eval WinDev2401Eval image and was able to deploy that from OVF. It was about 25GB. No problems with it.

I'm hoping I have no further issues. I'll let it run a bit. If all goes well I plan on joining the VMUG Advantage program.

Thanks,
David