Default MTU settings are not consistent

Summary

CAPM3 RKE2 cluster MTU settings are not consistent by default

Steps to reproduce

  1. Deploy a CAPM3 RKE2 cluster
  2. Observe network interfaces on the node
  3. Observe calico-node Pods logs

What is the current bug behavior?

The bare metal node network interfaces show a MTU of 1450.
vxlan.calico interface show a MTU of 1400.
All calico-nodes are pushing an error log every 10 seconds :

2024-10-07 15:55:38.258 [INFO][55] felix/vxlan_mgr.go 680: VXLAN device MTU needs to be updated device="vxlan.calico" ipVersion=0x4 new=1450 old=1400
2024-10-07 15:55:38.258 [WARNING][55] felix/vxlan_mgr.go 682: Failed to set vxlan tunnel device MTU error=invalid argument ipVersion=0x4
2024-10-07 15:55:48.262 [INFO][55] felix/vxlan_mgr.go 680: VXLAN device MTU needs to be updated device="vxlan.calico" ipVersion=0x4 new=1450 old=1400
2024-10-07 15:55:48.263 [WARNING][55] felix/vxlan_mgr.go 682: Failed to set vxlan tunnel device MTU error=invalid argument ipVersion=0x4
2024-10-07 15:55:58.267 [INFO][55] felix/vxlan_mgr.go 680: VXLAN device MTU needs to be updated device="vxlan.calico" ipVersion=0x4 new=1450 old=1400
2024-10-07 15:55:58.267 [WARNING][55] felix/vxlan_mgr.go 682: Failed to set vxlan tunnel device MTU error=invalid argument ipVersion=0x4

What is the expected correct behavior?

  1. Interfaces MTU should default to 1500 and vxlan.calico should have 1450 MTU due to VXLAN overhead.
  2. No MTU error should be reported by calico-node Pods.