devices(ubiquiti): the Dream Machines' SFP+ cages carry the controller's WAN2/LAN #2369

Merged
jaylouisw merged 1 commit from catalogue/1121-0452 into main 2026-11-21 05:08:56 +00:00
Owner

Refs #130 (no closing keyword: the Dream Wall Pro still has no vendor page - techspecs.ui.com/unifi/cloud-gateways/udw-pro answers 307 with Location: /, re-measured 2026-11-21 04:5x UTC - so this PR does not close the request and must not).

UDM-Pro, UDM-Pro-Max and UDM-SE shipped both 10G SFP+ cages with no role, so ownsIpOn's separation (owner.role && inIface.role && owner.role !== inIface.role) never fired for them: with 192.168.1.1/24 on port1, an address held on the LAN bank was answerable on a cage.

The old note justified the omission with "the datasheet prints one LAN and one WAN cage without numbering them". True of the datasheet, false of the controller - and the count closes exactly:

surface measurement (2026-11-21)
https://static.ui.com/fingerprint/ui/public.json (200, 832,585 B), sku = UDM-Pro / UDM-Pro-Max / UDM-SE numberOfPorts 11 = standard 9 + plus [10, 11]; linkNegotiation eth9 -> portIdx 10 (10000/1000 FDX), eth10 -> portIdx 11 (10000/1000 FDX); ports names eth9 "WAN2" and eth10 "LAN"
https://techspecs.ui.com/unifi/cloud-gateways/udm-pro (200, 676,527 B) Port Layout 1 GbE RJ45 9 (1G/100M/10M) / 10G SFP+ 2 (10G/1G), Default WAN Ports (1) 10G SFP+ (1) GbE RJ45, Max. WAN Port Count 8
https://techspecs.ui.com/unifi/cloud-gateways/udm-pro-max (200, 677,375 B) same rows with 2.5 GbE RJ45 1
https://techspecs.ui.com/unifi/cloud-gateways/udm-se (200, 682,320 B) same rows with 2.5 GbE RJ45 1
https://dl.ui.com/ds/udm-pro.pdf (200, 353,553 B, 2 pp, text layer) SFP+ interface (1) LAN: 10G SFP+ / (1) WAN: 10G SFP+

Nine RJ45 positions plus two cage positions is the whole box, so the cages are positions 10 and 11 and the first SFP+ is the default second WAN in panel order: sfp1 / SFP+ 1 now role "wan", sfp2 / SFP+ 2 "lan", both keeping dualRole ["wan","lan"] because UniFi reassigns these cages and Max. WAN Port Count 8 is a ceiling.

udw.json is untouched on purpose: the same database prints standard 19 beside plus [17, 20] on 20 sockets there, i.e. 21 positions for a 20-port box, so its positions do not close and its cage order cannot be placed - recorded in the spec note so the next pass does not read the omission as an oversight.

Gate: node scripts/syntax-check.js 515 files OK; node scripts/audit-port-types.js 0 gated; node scripts/regen-registry-doc.js catalogue pages current (2,351 models, no diff); hardware-invariants 26, importer-roundtrip 26, port-roles 19, port-roundtrip 45, catalogue-hygiene 25, doc-catalog-counts 43, faceplate-catalogue 14 all pass; test/ubiquiti-wan-lan-role-separation.test.js 162 passed, 0 failed (new positives: each model's WAN cage refuses the LAN address, its LAN cage still serves it).

Refs #130 (no closing keyword: the Dream Wall Pro still has no vendor page - `techspecs.ui.com/unifi/cloud-gateways/udw-pro` answers 307 with `Location: /`, re-measured 2026-11-21 04:5x UTC - so this PR does not close the request and must not). `UDM-Pro`, `UDM-Pro-Max` and `UDM-SE` shipped both 10G SFP+ cages with no `role`, so `ownsIpOn`'s separation (`owner.role && inIface.role && owner.role !== inIface.role`) never fired for them: with `192.168.1.1/24` on `port1`, an address held on the LAN bank was answerable on a cage. The old note justified the omission with "the datasheet prints one LAN and one WAN cage without numbering them". True of the datasheet, false of the controller - and the count closes exactly: | surface | measurement (2026-11-21) | |---|---| | `https://static.ui.com/fingerprint/ui/public.json` (200, 832,585 B), `sku` = UDM-Pro / UDM-Pro-Max / UDM-SE | `numberOfPorts` 11 = `standard` 9 + `plus` [10, 11]; `linkNegotiation` `eth9` -> portIdx 10 (10000/1000 FDX), `eth10` -> portIdx 11 (10000/1000 FDX); `ports` names `eth9` "WAN2" and `eth10` "LAN" | | `https://techspecs.ui.com/unifi/cloud-gateways/udm-pro` (200, 676,527 B) | `Port Layout 1 GbE RJ45 9 (1G/100M/10M) / 10G SFP+ 2 (10G/1G)`, `Default WAN Ports (1) 10G SFP+ (1) GbE RJ45`, `Max. WAN Port Count 8` | | `https://techspecs.ui.com/unifi/cloud-gateways/udm-pro-max` (200, 677,375 B) | same rows with `2.5 GbE RJ45 1` | | `https://techspecs.ui.com/unifi/cloud-gateways/udm-se` (200, 682,320 B) | same rows with `2.5 GbE RJ45 1` | | `https://dl.ui.com/ds/udm-pro.pdf` (200, 353,553 B, 2 pp, text layer) | `SFP+ interface (1) LAN: 10G SFP+` / `(1) WAN: 10G SFP+` | Nine RJ45 positions plus two cage positions is the whole box, so the cages are positions 10 and 11 and the first SFP+ is the default second WAN in panel order: `sfp1` / `SFP+ 1` now `role` "wan", `sfp2` / `SFP+ 2` "lan", both keeping `dualRole` ["wan","lan"] because UniFi reassigns these cages and `Max. WAN Port Count 8` is a ceiling. `udw.json` is untouched on purpose: the same database prints `standard` 19 beside `plus` [17, 20] on 20 sockets there, i.e. 21 positions for a 20-port box, so its positions do not close and its cage order cannot be placed - recorded in the spec note so the next pass does not read the omission as an oversight. Gate: `node scripts/syntax-check.js` 515 files OK; `node scripts/audit-port-types.js` 0 gated; `node scripts/regen-registry-doc.js` catalogue pages current (2,351 models, no diff); `hardware-invariants` 26, `importer-roundtrip` 26, `port-roles` 19, `port-roundtrip` 45, `catalogue-hygiene` 25, `doc-catalog-counts` 43, `faceplate-catalogue` 14 all pass; `test/ubiquiti-wan-lan-role-separation.test.js` 162 passed, 0 failed (new positives: each model's WAN cage refuses the LAN address, its LAN cage still serves it).
devices(ubiquiti): the Dream Machines' SFP+ cages carry the controller's WAN2/LAN (refs #130)
All checks were successful
CI / Classify the change (pull_request) Successful in 15s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 15s
CI / Test (Node 22) (pull_request) Successful in 1m1s
CI / Server smoke and Docker build (pull_request) Successful in 27s
CI / Test (Node 18) (pull_request) Successful in 1m4s
CI / Test (Node 20) (pull_request) Successful in 1m4s
CI / Arm auto-merge (pull_request) Successful in 4s
3c1a89831e
UDM-Pro, UDM-Pro-Max and UDM-SE shipped both 10G SFP+ cages with no role,
so ownsIpOn's separation (owner.role && inIface.role && owner.role !==
inIface.role) did not fire on them: an address held on port1 was answered
on the cage too.

The datasheet numbers the cages only as counts, but the controller database
does number them: numberOfPorts 11 = standard 9 + plus [10, 11], eth9 at
portIdx 10 is "WAN2", eth10 at portIdx 11 is "LAN". First cage = default
second WAN, second = LAN; dualRole keeps them reassignable.

udw.json is deliberately untouched: there standard 19 + plus [17, 20] on 20
sockets does not close, so its cage order cannot be placed.
forgejo-actions scheduled this pull request to auto merge when all checks succeed 2026-11-21 04:57:45 +00:00
jaylouisw deleted branch catalogue/1121-0452 2026-11-21 05:08:58 +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!2369
No description provided.