Kubernetes Gateway API: Separating Infrastructure, Routing, and Application Ownership
How GatewayClass, Gateway, and typed Route resources improve multi-team traffic management beyond a single overloaded Ingress object.
tag
29 posts
How GatewayClass, Gateway, and typed Route resources improve multi-team traffic management beyond a single overloaded Ingress object.
A layered account of Haiku networking, from device drivers and kernel protocols through routes, DNS, services, and application APIs.
Reconstruct the October 29, 1969 UCLA-to-SRI login from the IMP log, and separate ARPANET's first host message from later Internet myths.
How Ethernet moved from a 2.94-megabit coax experiment to switched full-duplex links without depending on collisions forever.
What WSL's automatic proxy discovery can transfer from Windows to Linux, where individual tools still diverge, and how to debug enterprise proxies safely.
How to correlate FreeBSD link events, counters, negotiated media, switch logs, optics, driver messages, and EEE before changing anything.
How to diagnose FreeBSD ntpd source selection, reachability, offset thresholds, drift, virtual clocks, and safe one-time clock steps.
How to configure modern FreeBSD CARP on two hosts, control election and preemption, test failover, synchronize PF state, and monitor HA.
How to build and verify multiple FreeBSD routing tables, assign services and jails to FIBs, use PF policy routing, and distinguish them from MPTCP.
Diagnose Haiku DHCP failures in order: device and link, Wi-Fi association, lease and routes, DNS, static comparison, logs, and reports.
Editing /etc/resolv.conf directly on a systemd-resolved system silently fails or reverts — the correct way to override DNS settings that persists.
Creating a separate network stack on the same machine to test firewall rules or networked software without touching the host's real network.
Working through the network preference files macOS updates most commonly leave inconsistent, before assuming a hardware or router problem.
Using the same BSD packet filter that powers macOS's Application Firewall directly, for filtering the GUI's simple allow/deny-per-app model can't express.
The layered, extensible packet-filtering architecture in Windows that both Windows Firewall and most third-party security software build on.
Working through the DNS client cache, adapter-specific resolver settings, and configured servers systematically, before jumping to a driver reinstall.
Isolating whether a WSL2 networking failure is DNS configuration, VPN interference, or the networking mode itself, before reaching for a fix.
Switching between WSL2's default NAT networking and mirrored networking mode, and verifying which one actually solves your specific reachability problem.
Combining two or more network interfaces into a single bonded interface with NetworkManager — for redundancy, throughput, or both, by mode.
WSL2's original NAT networking isolates it from your LAN by default; mirrored mode, added later, takes a fundamentally different approach with real tradeoffs.
How pod-to-pod networking, Services, and kube-proxy's packet rewriting fit together to make Kubernetes' flat network model actually work.
Getting real TCP/IP networking working on FreeDOS using mTCP and a packet driver, enough for FTP, Telnet, and a basic text-based browser.
A hop-by-hop VNET jail diagnostic for epair ownership, addresses, routes, bridges or NAT, forwarding, PF, DNS, and misleading ping failures.
How to trace a hung FreeBSD NFS operation through process waits, protocol version, transport, server health, firewall paths, and mount policy.
A rollback-safe FreeBSD static-network workflow for interface identity, IPv4 and IPv6 addressing, routes, resolver ownership, staged activation, and validation.
A least-privilege FreeBSD WireGuard setup covering if_wg, protected keys, cryptokey routing, PF exposure, NAT peers, rc startup, traffic tests, and rotation.
How FreeBSD exposes interfaces, addresses, routes, FIBs, bridges, VLANs, link aggregation, sockets, and packet-level evidence.
How FreeBSD PF evaluates rules, creates state, uses tables and anchors, performs NAT, logs decisions, and supports safe remote changes.
Ping by IP works but hostnames don't resolve — a systematic path through resolv.conf, systemd-resolved, and nsswitch.conf finds where it breaks.