High availability of VF-based ROCE protocol



I successfully configured roce over lag with ”https://enterprise-support.nvidia.com/s/article/How-to-Configure-RoCE-over-LAG-ConnectX-4-ConnectX-5-ConnectX-6“.
Is it only support PF? Is there any solution that supports for VF?
I want to support roce communication in k8s containers and need a VF-based high availability solution.

Best regards,


You might want to take a look at this link below, applicable to CX6 as well.
I believed this is what you are looking for.

Kubernetes Using SR-IOV

In order to use RDMA in Kubernetes environment with SR-IOV networking mode, two main
components are required:
RDMA device plugin - this plugin allows for exposing RDMA devices in a Pod
SR-IOV CNI plugin - this plugin provisions VF net device in a Pod


Thanks a lot for your answer. I have already configured it use that way. But I need a solution that is high available for VF, likes NIC bond or Roce LAG.

Maybe check ASAP² Direct solution, information can be retrieved from our MLNX_OFED driver UM or we have some articles on our community site.

1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.