Hi everyone,
I’m working with an NVIDIA SN5400 switch that comes preloaded with Cumulus Linux, which is great.
For our project, we’re also using the NVIDIA Switch SDK on the same device. Our goal is to dynamically add new routing rules in real time based on the contents of UDP payloads — hence the need to use the SDK.
Specifically, we want to create VLANs and assign ports to them on the fly, according to the expected incoming traffic patterns. This setup worked perfectly on the SDK Developer OS. However, once NVUE is introduced, the dynamic VLAN creation no longer functions as expected.
When the NVUE service is disabled, SDK-based VLAN routing works again, but only for VLANs that were initially created through NVUE — which defeats the purpose of managing everything via the SDK. In this configuration, VLANs created by the SDK appear within the SDK environment but have no actual effect on the system.
Questions:
-
How can VLANs created via the SDK be made valid and active on the default bridge?
-
(Bonus) Is there a way to make NVUE and the SDK coexist and operate seamlessly together?