Hyper-V Teaming CPU issue

Question

Hello.

When using teaming solution (LBFO or SET), The CPU 0 usage increase until make crash the network (HV & VMs) and freeze the server.

Server :

Windows Server 2022 + HV : Fully updated

HP ML350 G10 + Latest SPP + Network Drivers

It's a team of 4 1Gb adapter

Progress :

With 1 adapter connected the CPU 0 go from 3% to 50%.

With 2 adapter connected the CPU 0 go from 50% to 80%.

With 3 adapter connected the CPU 0 go from 80% to 100%, the latency increase from 1 ms to 3500ms to finally stop any communication.

With 4 adapter connected, that just appear faster.

Physically the HV server completely freeze, no action possible.

Remediation :

By disconnecting 3 port on the switch, the server come back to life like nothing happened.

Advices?

Thx.

Answer 1

Thx for your reply.

The problem is solved.

It was network spam (not a broadcast storm, from a powered off device :-( ), which is not a problem for a network card, but for a team of 4 it caused a crash.

Answer 2

Hello there,

Have you checked the Event Viewer for any trace?

If you cannot find anything in Event Viewer I would suggest you use tools like Procmon to find the root cause for this behaviour.

Review your NIC teaming configuration settings. Make sure that the teaming mode is set correctly for your environment (e.g., switch independent vs. switch dependent). Also verify that you have selected the appropriate load balancing algorithm (e.g., address hash, Hyper-V port) based on your needs.

Similar discussion here https://learn.microsoft.com/en-us/answers/questions/828128/hyper-v-problem-after-add-10-vms?orderby=oldest

Hope this resolves your Query !!

--If the reply is helpful, please Upvote and Accept it as an answer–