I feel like the title is somewhat appropriate: part of the post was about selectively removing parts of Kubernetes networking, and digging through the kernel networking stack, to troubleshoot the issue