OpenWrt Two Approval

Original Article ↗ Hacker News Discussion ↗

What OpenWrt Two Is and Pricing

  • Several commenters were initially confused; “Two” is a dedicated hardware router (successor to OpenWrt One), not a software release.
  • Target price is “around $250,” with part of the price going back to the project.
  • Some only learned One existed while trying to understand Two, suggesting communication/marketing from the project has been weak.

Port and SoC Design Debates

  • Major discussion around the choice of 5GBASE‑T copper vs additional 10G SFP+ ports.
  • Pro–5G copper: works over existing Cat5e/Cat6 cabling, easy home upgrade path, better fit for consumer market, and 5G can be a power/heat sweet spot when uplinking to 10G switches.
  • Pro–extra SFP+: more flexible (1/2.5/5/10G, copper or fiber), better for advanced/home-lab and professional setups; some feel a fixed 5G port is a “lock‑in.”
  • Realtek 5G PHY is highlighted because it avoids firmware blobs and helps make 2.5/5G mainstream at lower cost.
  • Some wish the design had at least dual 10G SFP+ and fewer 1G ports, but others point out SoC lane/topology constraints limit this.

Performance and QoS

  • Concerns whether Two’s CPU will be fast enough to run CAKE SQM at line rate, given reports that One cannot.
  • MediaTek hardware fq_codel/HQoS exists but is noted as not supported in vanilla OpenWrt, only in vendor SDKs.

Partner Choice: GL.iNet and Maintenance

  • Many are enthusiastic: GL.iNet hardware is praised as performant and good value; users report better real‑world Wi‑Fi than some “premium” gear.
  • Others criticize GL.iNet’s historic pattern: old kernels, messy or poorly maintained sources, OpenWrt presented as an “advanced” unsupported mode, and a push toward cloud‑managed stock firmware.
  • Hope that a first‑class “OpenWrt Two” collaboration will force better upstreaming and long‑term maintenance.

Security, Jurisdiction, and Supply‑Chain Trust

  • Long sub‑thread on trusting a Hong Kong/Shenzhen‑linked vendor in a security‑sensitive role.
  • Some are broadly suspicious of Chinese hardware (citing Huawei and consumer PC experiences); others argue this is overblown and that US vendors and agencies are hardly more trustworthy.
  • Nuanced view:
    • Consumer barebones PCs and OpenWrt‑flashed routers have limited realistic attack surface in firmware compared to complex telco gear.
    • Firmware rarely updated is a problem, but not always a deal‑breaker; running current OpenWrt on older firmware is still safer than stock ISP/consumer routers.
  • Several point out it’s nearly impossible to avoid Chinese manufacturing entirely; alternatives like MikroTik, ODROID, Synology, QNAP, and various white‑box boards are mentioned but often cost more or have their own issues (closed source, GPL friction, weak Wi‑Fi).

Project Governance and the Vote

  • The formal vote result (24 yes, 0 no, 18 missing) is seen as effectively unanimous approval, with “missing” read as abstentions rather than opposition.
  • Some question the point of a yes/no vote when there’s little real internal contention; it’s described as functioning more like a project‑lead or project‑greenlight ritual than a true conflict resolution mechanism.
  • Hypothetical reasons for a “no” (budget, unreliable leadership, bad contractor choice) are discussed as what might matter in other circumstances.

Critiques of OpenWrt’s Overall Direction

  • One commenter provides a detailed indictment of OpenWrt’s trajectory:
    • Claims core bugs, bugfixes, and routing improvements are neglected; some subsystems allegedly aren’t open to normal PRs/bug reports.
    • Complaints about old devices being abandoned when kernels no longer fit flash, with no accommodation for older kernels.
    • Frustration that effort is going into GPU/video acceleration and desktop‑style stacks (Mesa, X/Wayland, Doom) instead of routing features and hardware acceleration.
  • They argue OpenWrt One (and likely Two) are e‑waste due to: fixed/limited RAM, non‑replaceable Wi‑Fi, lack of expansion (M.2/mini‑PCIe, SATA), and narrow use as “just a router.”
  • Counter‑arguments:
    • Highly integrated router SoCs are cheaper and closer to what mass‑market vendors ship; improving support there benefits the broader ecosystem.
    • Expecting router SoCs with SATA/eSATA or modular radios is unrealistic given current vendor priorities; PCIe lanes and packaging are tuned for all‑in‑one Wi‑Fi routers.
    • Focusing on PC‑style modularity would push OpenWrt further into “general Linux distro” territory and promote exactly the distractions critics dislike.

Comparisons and Alternative Visions

  • Banana Pi R4 and similar multi‑10G boards are cited as higher‑spec/cheaper competitors; some feel $250 is steep relative to such offerings, others say R4 has its own compromises (e.g., weaker radios).
  • Turris Omnia is held up as a counterexample: despite being almost 10 years old, replaceable Wi‑Fi, more RAM, and expansion give it long life as router + NAS + home server.
  • A separate thread dreams about a single box that is router, media hub, smart speaker, smart‑home bridge, charger, NAS, and VPN/DNS server; replies suggest cobbling this together from Synology, OpenWrt + storage, or small PCs with multiple NICs rather than expecting it from a focused router design.

Sentiment Summary

  • Many are excited: they see OpenWrt Two as strong, open, Wi‑Fi 7‑capable hardware at a fair price that directly benefits the project.
  • Skeptics worry about: hardware longevity (non‑modular design), GL.iNet’s history, OpenWrt’s strategic focus, and geopolitical/supply‑chain risk.
  • Overall tone: cautiously optimistic enthusiasm, with a sizable contingent arguing for more modular, longer‑lived, and routing‑centric designs.