devices(ubiquiti): the Fiber pair's copper bank served its LAN address on the uplink (#130) #2380

Merged
jaylouisw merged 1 commit from catalogue/1121-0622 into main 2026-11-21 06:43:55 +00:00
Owner

Refs #130 (no closing keyword - the Dream Wall Pro itself is still absent from every Ubiquiti surface; this is the adjacent increment from the same thread).

What changed

UCG-Fiber and UXG-Fiber shipped their entire copper bank with no role, so ownsIpOn's separation (owner.role && inIface.role && owner.role !== inIface.role, engine/device.js) never fired. Measured on the pre-change specs: with 192.168.1.1/24 on the first LAN port, ownsIpOn(WAN, '192.168.1.1') returned true on UCG-Fiber and ownsIpOn(eth4, '192.168.1.1') returned true on UXG-Fiber — a host cabled to the 10G RJ45 uplink inside the LAN prefix was answered by the LAN gateway.

Both rows now carry the vendor's split:

row LAN bank uplink source
devices/ubiquiti/ucg-fiber.json Port 1-Port 4 -> role: lan WAN -> role: wan controller ports map + tech-spec
devices/ubiquiti/uxg-fiber.json eth0-eth3 -> role: lan eth4 -> role: wan controller ports map (names already match) + tech-spec

Evidence (read 2026-11-21)

https://static.ui.com/fingerprint/ui/public.json — HTTP 200, 832,585 B, sku = UCG-Fiber / UXG-Fiber, both publish numberOfPorts 7 and ports = {eth0: LAN, eth1: LAN, eth2: LAN, eth3: LAN, eth4: WAN, eth5: LAN, eth6: WAN2}; linkNegotiation puts eth0-eth3 at portIdx 1-4 (2500/1000/100/10 FDX), eth4 at portIdx 5 (10000/5000/2500/1000/100/10 FDX) and the two cages at portIdx 6-7 (10000/1000 FDX only).

https://techspecs.ui.com/unifi/cloud-gateways/ucg-fiber — HTTP 200, 667,242 B: Port Layout 2.5 GbE RJ45 4 / 10 GbE RJ45 1 / 10G SFP+ 2, Default WAN Ports (1) 10G SFP+ (1) 10 GbE RJ45, Max. WAN Port Count 6.

https://techspecs.ui.com/unifi/cloud-gateways/uxg-fiber — HTTP 200, 649,433 B: same rows.

What is deliberately not modelled

The two SFP+ cages on each row stay role-less. The controller does number them (eth5 LAN, eth6 WAN2), but neither row's cage names (SFP+ 1/SFP+ 2, SFP+-1/SFP+-2) are vendor-issued — both pages print one undifferentiated 10G SFP+ 2 row — so assigning a role would be an ordering guess that moves a lab's saved uplink. Max WAN Port Count 6 stays unmodelled for the same reason. Both notes record this.

Gates

  • node scripts/syntax-check.js -> 516 files, all OK
  • node scripts/audit-port-types.js -> gated 0 contradictions
  • node scripts/audit-devices.js -> 2,352 models / 29,768 sockets, 0 uncablable, 0 hard defects
  • node scripts/check-control-bytes.js -> clean
  • node scripts/run-tests.js -> 436 suites, 14,678 assertions, 0 failed (includes the 14 new assertions in test/ubiquiti-wan-lan-role-separation.test.js)
  • node scripts/regen-registry-doc.js -> catalogue pages already current
Refs #130 (no closing keyword - the Dream Wall Pro itself is still absent from every Ubiquiti surface; this is the adjacent increment from the same thread). ## What changed `UCG-Fiber` and `UXG-Fiber` shipped their entire copper bank with no `role`, so `ownsIpOn`'s separation (`owner.role && inIface.role && owner.role !== inIface.role`, engine/device.js) never fired. Measured on the pre-change specs: with `192.168.1.1/24` on the first LAN port, `ownsIpOn(WAN, '192.168.1.1')` returned **true** on UCG-Fiber and `ownsIpOn(eth4, '192.168.1.1')` returned **true** on UXG-Fiber — a host cabled to the 10G RJ45 uplink inside the LAN prefix was answered by the LAN gateway. Both rows now carry the vendor's split: | row | LAN bank | uplink | source | |---|---|---|---| | `devices/ubiquiti/ucg-fiber.json` | `Port 1`-`Port 4` -> `role: lan` | `WAN` -> `role: wan` | controller `ports` map + tech-spec | | `devices/ubiquiti/uxg-fiber.json` | `eth0`-`eth3` -> `role: lan` | `eth4` -> `role: wan` | controller `ports` map (names already match) + tech-spec | ## Evidence (read 2026-11-21) `https://static.ui.com/fingerprint/ui/public.json` — HTTP 200, 832,585 B, `sku` = UCG-Fiber / UXG-Fiber, both publish `numberOfPorts` 7 and `ports` = `{eth0: LAN, eth1: LAN, eth2: LAN, eth3: LAN, eth4: WAN, eth5: LAN, eth6: WAN2}`; `linkNegotiation` puts eth0-eth3 at portIdx 1-4 (2500/1000/100/10 FDX), eth4 at portIdx 5 (10000/5000/2500/1000/100/10 FDX) and the two cages at portIdx 6-7 (10000/1000 FDX only). `https://techspecs.ui.com/unifi/cloud-gateways/ucg-fiber` — HTTP 200, 667,242 B: `Port Layout 2.5 GbE RJ45 4 / 10 GbE RJ45 1 / 10G SFP+ 2`, `Default WAN Ports (1) 10G SFP+ (1) 10 GbE RJ45`, `Max. WAN Port Count 6`. `https://techspecs.ui.com/unifi/cloud-gateways/uxg-fiber` — HTTP 200, 649,433 B: same rows. ## What is deliberately not modelled The two SFP+ cages on each row stay role-less. The controller does number them (eth5 LAN, eth6 WAN2), but neither row's cage names (`SFP+ 1`/`SFP+ 2`, `SFP+-1`/`SFP+-2`) are vendor-issued — both pages print one undifferentiated `10G SFP+ 2` row — so assigning a role would be an ordering guess that moves a lab's saved uplink. `Max WAN Port Count 6` stays unmodelled for the same reason. Both notes record this. ## Gates - `node scripts/syntax-check.js` -> 516 files, all OK - `node scripts/audit-port-types.js` -> gated 0 contradictions - `node scripts/audit-devices.js` -> 2,352 models / 29,768 sockets, 0 uncablable, 0 hard defects - `node scripts/check-control-bytes.js` -> clean - `node scripts/run-tests.js` -> 436 suites, 14,678 assertions, 0 failed (includes the 14 new assertions in `test/ubiquiti-wan-lan-role-separation.test.js`) - `node scripts/regen-registry-doc.js` -> catalogue pages already current
devices(ubiquiti): the Fiber pair's copper bank answered its LAN address on the uplink (#130)
All checks were successful
CI / Classify the change (pull_request) Successful in 20s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 22s
CI / Test (Node 22) (pull_request) Successful in 1m4s
CI / Test (Node 18) (pull_request) Successful in 1m5s
CI / Test (Node 20) (pull_request) Successful in 1m7s
CI / Server smoke and Docker build (pull_request) Successful in 25s
CI / Arm auto-merge (pull_request) Successful in 5s
736aaf13e1
forgejo-actions scheduled this pull request to auto merge when all checks succeed 2026-11-21 06:34:12 +00:00
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
jaylouisw/packetforge!2380
No description provided.