The configuration of Calico network when joining a Windows node to the Kubernetes cluster is not effective.

[DongWli]

I first configured a Kubernetes Linux cluster and then added a Windows worker node according to the documentation.

Official documentation: https://docs.tigera.io/calico/latest/getting-started/kubernetes/windows-calico/requirements

Under the title "Linux platform requirements," I used the command kubectl patch ippool default-ipv4-ippool -p '{"spec":{"ipipMode":"Never","vxlanMode":"Always"}}'. During two validation processes, the first time I disabled the BGP network mode, and the command did not take effect. Later, I switched to the BGP network and modified the command to kubectl patch ippool default-ipv4-ippool -p '{"spec":{"ipipMode":"Never","vxlanMode":"Never"}}', but this command also did not work.

However, by directly modifying spec.encapsulation: None through kubectl edit installation default -o yaml -n tigera-operator, I was able to see that the network information changed after running kubectl get ippool default-ipv4-ippool -o yaml. This issue has been resolved, and I am only providing this test record as feedback.

[caseydavenport]

Thanks for raising this - sounds like our documentation still asks users to modify the IP pools directly, instead of modifying the operator Installation yaml.

This will vary based on install method - for operator-based installations, IP pools should be edited via the Installation.

For manifest based installs, you will need to modify the IP pools directly.