Virtualization Related Troubleshooting

Issue

Cause

Solution

Mellanox driver fails to load a host machine in SR-IOV environment and appears with yellow bang in Device Manager.

The device may not have been able to find enough free resources that it can use. (Code 12).

  1. Boot to BIOS and disable SR-IOV.

  2. Burn Firmware with lower number of VFs.

  3. Re-enable SR-IOV in BIOS.

For more information, please contact Mellanox support.

Running Windows server 2012, 2012 R2 and 2016 as VM over ESX with Mellanox adpter cards connected as Direct pass-through fails to power on.

ConnectX adapter network cards might be trying to use too many MSI-X vectors.

  1. Go to the vSphere Web Client.

  2. Right-click the virtual machine and select Edit Settings.

  3. Click the Options tab and expand Advanced.

  4. Click Edit Configuration.

  5. Click Add Row.

  6. Add the parameter to the new row:

  • In the Name column, add pciPassthru0.maxMSIXvectors.

  • In the Value column, add 31.

7. Click OK and click OK again.

For further details, please refer to: http://kb.vmware.com/selfservice/microsites/search.do?cmd=displayKC&doc-Type=kc&externalId=2032981&sliceId=1& docTypeID=DT_KB_1_1&dialogID=408420191&stateId=1 0 388456420

When enabling the VMQ, in case NVGRE offload is enabled, and a teaming of two virtual ports is performed, no ping is detected between the VMs and/ or ping is detected but no establishing of TCP connection is possible.

Might be missing critical Microsoft updates.

Please refer to: http://support.microsoft.com/kb/2975719

“August 2014 update rollup for Windows server RT 8.1, Windows server 8.1, and Windows server 2012 R2” – specifically, fixes.

The VF adapter is in a 'yellow bang' state, and the following message appears in the host event viewer massage log: "Self-Healing second tier policy was activated on Virtual Function (VF) #%3 with sensors #%4, and Bus driver restart is needed on the VF. Please restart the Bus driver."

A self-healing reset was activated on the VF, but the driver is in a yellow bang' state and must be manually restarted in order to continue the operation.

Restart the VF driver.

In Hyper-V environment, Enable-Net-AdapterVmq powershell command can enable VMQ on a network adapter only if the virtual switch which does not have SR-IOV enabled is defined over corresponding network adapter.

The powershell command might depend on two registry fields: *VMQ and

*RssOrVmqPreference, when the former is controlled by powershell and the latter is controlled by the virtual switch.

For further information on these registry keys, please refer to: http://msdn.microsoft.com/en-us/library/windows/hardware/ hh451362(v=vs.85).aspx

Mellanox driver fails to load in a guest machine in SR-IOV environment and appears with yellow bang in the Device Manager.

The host machine cannot reserve enough QPs for the specific VF.

Increase the LogNumQP in the mlx4_bus registry.

Mellanox driver fails to load in a guest machine with Windows 10 Inbox driver version 4.91 in SR-IOV environment and appears with yellow bang in the Device Manager.

The host machine cannot reserve enough QPs for the specific VF.

Update to the latest version of Windows 10 or install driver version 5.22 and above.

© Copyright 2023, NVIDIA. Last updated on Oct 26, 2023.