Hello,
Quick overview of my Horizon View cluster environment which contains 4 identical host:
UCSC-C240-M5SX
(2) NVIDIA Tesla T4 GPU in each host
ESXi - 7.0.1, 17168206
ESXi VIB from NVIDA - 450.89
I am able to configure a VM with vGPU resources and when I run "nvidia-smi" I see the allocation and all appears to be working as expected inside of the VM. However, when I attempt a live vmotion I get the following error:
One or more devices (pciPassthru0) required by VM are not available on host.
If I shutdown the VM, move it and power it on everything is fine. I am seeing that the X.Org service is crashing on the host I am trying to move the VM to after a few minutes of it running. If that service remains running (which its not), I am able to provide live vmotion of VMs with vgpu resources assigned.
I noticed that the X.Org service is stopped and will not start on the other host in the cluster as well. I have tried to uninstall and reinstall the VIB which has not helped. I opened a case with VMware and the tech said all the logs I sent look good. We have a session tomorrow to further troubleshoot but I am at a loss right now. In the past, when the X.Org service would not start, it was related to the VIB and it needing to be re-installed.
If anyone is running a similar configuration I am interested in any support that can be provided.
Thank you
Hi ITVisionIT,
Where you able to address this issue?
Please advise.
I have the same error while trying to vmotion a VDI that has a vgpu attached. Did you ever solve this?