Calico v3.20 nodes don't become ready

[CecileRobertMichon]

/kind bug

What steps did you take and what happened:
[A clear and concise description of what the bug is.]

What did you expect to happen: Tried to update Calico to v3.20.0 in #1583, cluster provisioning failed because nodes didn't become ready. Not sure if there is something wrong with Calico v3.20 + vxlan, or if a configuration changed in the manifest which caused it to fail.

To repro, build a cluster and apply https://docs.projectcalico.org/v3.20/manifests/calico-vxlan.yaml

We should investigate what the error is and open an issue on the Calico repo if there is a regression in 3.20.

Anything else you would like to add:
[Miscellaneous information that will assist in solving the issue.]

Environment:

• cluster-api-provider-azure version:
• Kubernetes version: (use kubectl version):
• OS (e.g. from /etc/os-release):

[CecileRobertMichon]

/assign @nader-ziada

[CecileRobertMichon]

it also failed with v3.19.2 when using https://docs.projectcalico.org/v3.19/manifests/calico.yaml so it might be due to something that changed in the manifest

the veth_mtu change is the first obvious one, might be worth exploring