[Device]: OptiXtrans E9605 #807

Closed
opened 2026-11-08 06:21:18 +00:00 by jaylouisw · 14 comments
Owner

Vendor

Huawei

Exact model name

OptiXtrans E9605

Category

Other / not sure

Ports and interfaces

  • 5 service slots
  • Subrack 177mm (H) x 442mm (W) x 295mm (D)
  • Line rate per channel up to 1.2 Tbit/s

PoE budget (watts)

None (no PoE output)

Datasheet or product page URL

https://e.huawei.com/en/products/optical-transmission/optixtrans-e9600

What do you want to model with it?

Compact edge optical transport (OTN/WDM) node.

Capabilities

OTN/WDM optical transport

Before submitting

  • I searched the catalog and existing issues, and this model is not there yet.
  • The port list above comes from a datasheet, not from memory or another wiki.
### Vendor Huawei ### Exact model name OptiXtrans E9605 ### Category Other / not sure ### Ports and interfaces - 5 service slots - Subrack 177mm (H) x 442mm (W) x 295mm (D) - Line rate per channel up to 1.2 Tbit/s ### PoE budget (watts) None (no PoE output) ### Datasheet or product page URL https://e.huawei.com/en/products/optical-transmission/optixtrans-e9600 ### What do you want to model with it? Compact edge optical transport (OTN/WDM) node. ### Capabilities OTN/WDM optical transport ### Before submitting - [x] I searched the catalog and existing issues, and this model is not there yet. - [x] The port list above comes from a datasheet, not from memory or another wiki.
Author
Owner

Closing status: blocked on a stated configuration.

The OptiXtrans E6600/E9600 specification tables publish what a frame can do - service board slots (E6600: "Number of Service Board Slots: DC chassis 14 / AC chassis 12"), cross-connect capacity, maximum wavelengths and per-channel rate - and nothing about ports, because an OTN subrack has none until service boards are fitted. A 400G/λ, 96-wavelength figure is a capacity, not a socket count.

So the issue stays open; docs/DEVICE_REGISTRY.md §"What is left open, and why" records the reason for the family (#803, #804, #807, #808, #810).

What would close it: the board set you intend to model (e.g. "E6616 with two TN13LSC boards"), or a decision to model the frame's slots as a repeatable module - which would be an engine feature, not a spec file.

Closing status: **blocked on a stated configuration.** The OptiXtrans E6600/E9600 specification tables publish what a *frame* can do - service board slots (E6600: "Number of Service Board Slots: DC chassis 14 / AC chassis 12"), cross-connect capacity, maximum wavelengths and per-channel rate - and nothing about ports, because an OTN subrack has none until service boards are fitted. A 400G/λ, 96-wavelength figure is a capacity, not a socket count. So the issue stays open; `docs/DEVICE_REGISTRY.md` §"What is left open, and why" records the reason for the family (#803, #804, #807, #808, #810). **What would close it:** the board set you intend to model (e.g. "E6616 with two TN13LSC boards"), or a decision to model the frame's slots as a repeatable module - which would be an engine feature, not a spec file.
Author
Owner

Re-checked against both current specification tables; still blocked, but the blocker is now exact and the frame turned out to be the harder of the two kinds.

Sources read on 2026-11-16:

What the tables settle. The E9605 is 5 service slots with Cross-connect Type: NA — like the E9612, it is a photonic shelf, not a small OTN cross-connect. So the earlier note ("an OTN subrack has no ports until service boards are fitted") understates this one: there is no chassis-only port to model at all, not even a management Ethernet port in the published table. Every cableable port an E9605 will ever have is the client side of an OTU board fitted in one of those five slots.

Exact missing field: the board part number per slot — e.g. OptiXtrans E9605 (5x <board>), the label shape docs/RESOLUTIONS.md already sets out. With that one line, the spec is slots x the board's own published port count and nothing is guessed.

What would unblock it, concretely: name the boards (Huawei part numbers) or point me at the Optical-Layer/OTU board datasheet for the E9600 series; the chassis arithmetic is already transcribed in research/optixtrans-slot-census.md (PR #1752), so it converts in one step. The same file covers #803, #804, #808 and #810, including the fact that E9624's 12/24 is 10/20 in 1:3 mode and that E6608T/E9612/E9605 all print NA for cross-connect.

This issue stays open — it is not being closed as unbuildable.

**Re-checked against both current specification tables; still blocked, but the blocker is now exact and the frame turned out to be the harder of the two kinds.** Sources read on 2026-11-16: - https://e.huawei.com/en/products/optical-transmission/optixtrans-e9600 — the E9600 spec table, E9624 / E9612 / E9605 columns. - https://e.huawei.com/en/products/optical-transmission/optixtrans-e6600 — the E6600 spec table, for the four E66xx frames in the same queue. - https://e.huawei.com/marketingcloud/pep/asset/20000001/Material/6c96e5825bc34cf58c30fd13e0d8e7f2/M3T1A669N1129732965963960451/FTTO%20Solution%20for%20Education%20Campus_K-12_-Brochure_25H1_.pdf — Huawei's own FTTO brochure, which prints the per-model slot counts and the E9624 mode split. - `EDOC1100624531` (OptiXtrans E9600 Board Datasheet) — HTTP 404 to a crawler user-agent, body `This document cannot be found.` **What the tables settle.** The E9605 is `5` service slots with `Cross-connect Type: NA` — like the E9612, it is a **photonic shelf**, not a small OTN cross-connect. So the earlier note ("an OTN subrack has no ports until service boards are fitted") understates this one: there is no chassis-only port to model at all, not even a management Ethernet port in the published table. Every cableable port an E9605 will ever have is the client side of an OTU board fitted in one of those five slots. **Exact missing field:** the board part number per slot — e.g. `OptiXtrans E9605 (5x <board>)`, the label shape `docs/RESOLUTIONS.md` already sets out. With that one line, the spec is `slots x the board's own published port count` and nothing is guessed. **What would unblock it, concretely:** name the boards (Huawei part numbers) or point me at the Optical-Layer/OTU board datasheet for the E9600 series; the chassis arithmetic is already transcribed in `research/optixtrans-slot-census.md` (PR #1752), so it converts in one step. The same file covers #803, #804, #808 and #810, including the fact that E9624's `12/24` is `10/20` in 1:3 mode and that E6608T/E9612/E9605 all print `NA` for cross-connect. This issue stays open — it is not being closed as unbuildable.
Author
Owner

The board census this issue asked for is public now; the blocker has moved off documentation and into the engine. PR #1788 records it (research/optixtrans-board-census.md); it does not close this issue.

Sources read 2026-11-17

  • https://e.huawei.com/marketingcloud/pep/asset/20000001/Material/0f9e00e3f2ca431693884a01667d7ec6/M3T1A669N1188254857740624027/OptiXtrans%20E9600%20Family%20-.pdf — HTTP 200, 277,705 B, vendor brochure: Max. number of service board slots | 1:1 cross-connect mode: 12 large slots or 24 small slots; 1:3 cross-connect mode: 10 large slots or 20 small slots (E9624) / 13 (E9612) / 5 (E9605). No board list.
  • https://www.actfornet.com/solutions/solution/OptiXtrans_E9600_Boards_Overview — HTTP 200, partner mirror of Huawei's board matrix: which board fits which subrack (lead, not evidence).
  • https://support.huawei.com/enterprise/en/optical-transmission/optixtrans-e9605-pid-250732995 — the real home of OptiXtrans E9600 Board Datasheet (2026-04-03), download behind the portal.
  • https://support.huawei.com/hedex/pages/EDOC1100126589BEJ09047/03/EDOC1100126589BEJ09047/03/resources/help/wdm/9800/hd/line/n210/n210.html — HTTP 403 to a crawler user-agent (536 B). This is the board description that would settle client-side vs WDM-side.
  • https://support.huawei.com/enterprise/en/doc/EDOC1100624531 — still 404, body This document cannot be found.

What the matrix settles for the E9605 (5 slots). It is narrower than the frame: OTN line boards TNG1N206 (6x10G), TNG1N210 (10x10G) and TNG1N401 (1x100G) fit, and so do the TNG1M* transponder units, TNG1A212, the mux/demux and ROADM boards. Every tributary board, every packet board, every TDM board and the turbo-WDM pair are NA for this subrack. Ten board rows, not forty. So the earlier note that the E9605 is "a photonic shelf" is right about cross-connect and wrong about service: it does have client ports, they just arrive through a transponder or a line board.

The exact missing field is therefore no longer the board part number — it is a port type. engine/media.js knows ethernet, wireless, console, usb, coax, dsl, powerline, serial, pon, caged as rj45 ... qsfpdd, xfp, cfp2 and fibre connectors. A TNG1N210 port is an OTU2/OTU2e WDM-side optical port (Huawei's published module list is SFP+ 10G, 10/40/80 km, DWDM 100 GHz and 50 GHz tunable, CWDM). Filed as type: ethernet, connector: sfp+ — the only value pair the engine accepts — that port becomes an Ethernet socket CAGE_MEDIA['sfp+'] will happily join to a 10G Ethernet peer. The cage is right, the protocol claim is wrong, and no vendor document can fix it.

What would unblock it: an optical-line port type (otn/wdm) in engine/media.js plus the faceplate renderer — a different stream. The moment one exists this becomes OptiXtrans E9605 (5x TNG1N210) = 50 x 10G OTU2 line ports, which is arithmetic on two published tables and nothing else. Until then the issue stays open and unmodelled, deliberately: a mislabelled port is worse than a missing device.

**The board census this issue asked for is public now; the blocker has moved off documentation and into the engine.** PR #1788 records it (`research/optixtrans-board-census.md`); it does not close this issue. **Sources read 2026-11-17** - `https://e.huawei.com/marketingcloud/pep/asset/20000001/Material/0f9e00e3f2ca431693884a01667d7ec6/M3T1A669N1188254857740624027/OptiXtrans%20E9600%20Family%20-.pdf` — HTTP 200, 277,705 B, vendor brochure: `Max. number of service board slots | 1:1 cross-connect mode: 12 large slots or 24 small slots; 1:3 cross-connect mode: 10 large slots or 20 small slots` (E9624) / `13` (E9612) / `5` (E9605). No board list. - `https://www.actfornet.com/solutions/solution/OptiXtrans_E9600_Boards_Overview` — HTTP 200, partner mirror of Huawei's board matrix: which board fits which subrack (lead, not evidence). - `https://support.huawei.com/enterprise/en/optical-transmission/optixtrans-e9605-pid-250732995` — the real home of `OptiXtrans E9600 Board Datasheet` (2026-04-03), download behind the portal. - `https://support.huawei.com/hedex/pages/EDOC1100126589BEJ09047/03/EDOC1100126589BEJ09047/03/resources/help/wdm/9800/hd/line/n210/n210.html` — **HTTP 403** to a crawler user-agent (536 B). This is the board description that would settle client-side vs WDM-side. - `https://support.huawei.com/enterprise/en/doc/EDOC1100624531` — still 404, body `This document cannot be found.` **What the matrix settles for the E9605 (5 slots).** It is narrower than the frame: OTN line boards `TNG1N206` (6x10G), `TNG1N210` (10x10G) and `TNG1N401` (1x100G) fit, and so do the `TNG1M*` transponder units, `TNG1A212`, the mux/demux and ROADM boards. Every **tributary** board, every packet board, every TDM board and the turbo-WDM pair are `NA` for this subrack. Ten board rows, not forty. So the earlier note that the E9605 is "a photonic shelf" is right about cross-connect and wrong about service: it does have client ports, they just arrive through a transponder or a line board. **The exact missing field is therefore no longer the board part number — it is a port type.** `engine/media.js` knows `ethernet, wireless, console, usb, coax, dsl, powerline, serial, pon`, caged as `rj45 ... qsfpdd`, `xfp`, `cfp2` and fibre connectors. A `TNG1N210` port is an OTU2/OTU2e WDM-side optical port (Huawei's published module list is SFP+ 10G, 10/40/80 km, DWDM 100 GHz and 50 GHz tunable, CWDM). Filed as `type: ethernet, connector: sfp+` — the only value pair the engine accepts — that port becomes an Ethernet socket `CAGE_MEDIA['sfp+']` will happily join to a 10G Ethernet peer. The cage is right, the protocol claim is wrong, and no vendor document can fix it. **What would unblock it:** an optical-line port type (`otn`/`wdm`) in `engine/media.js` plus the faceplate renderer — a different stream. The moment one exists this becomes `OptiXtrans E9605 (5x TNG1N210)` = 50 x 10G OTU2 line ports, which is arithmetic on two published tables and nothing else. Until then the issue stays open and unmodelled, deliberately: a mislabelled port is worse than a missing device.
Author
Owner

The board census now has a vendor URL, a document ID and a 404. And the second blocker on this family was wrong.

1. Huawei does publish a board datasheet, and the E9605 page links it. The E9605/E9612/E9624 support pages list, under Datasheet, OptiXtrans E9600 Board Datasheet -> /enterprise/en/doc/EDOC1100163901?idPath=24030884|9856748|9858640|250707904|250732995 (https://support.huawei.com/enterprise/en/optical-transmission/optixtrans-e9605-pid-250732995, 200, 436,367 bytes, crawler UA, read 2026-11-17). So the matrix is not "unpublished"; it is one document, and that document is withheld from anonymous readers:

  • /enterprise/en/doc/EDOC1100163901?idPath=... -> 302 -> https://support.huawei.com/view/notFound, 404.
  • /enterprise/ru/doc/EDOC1100163901?idPath=... -> 404.
  • /enterprise/en/datasheet/optixtrans-e9600-board-datasheet-pid-250732995 -> 200, but the body is the product page again (slug ignored, <title>Huawei OptiXtrans E9605 Support Guide...).
  • hedex/hdx.do?docid=EDOC1100163901&lang=en -> 200, 3,728 bytes, the SPA shell; &format=json returns the identical HTML.
  • Current generation fails the same way: EDOC1100625542 (V100R026C00 Hardware Description), EDOC1100622610 (Product Description), EDOC1100622612 (Feature Description) -> all 404 -> view/notFound.

Control, so this is not read as "the crawler route broke": https://support.huawei.com/enterprise/en/doc/EDOC1100318733/4f560405/cabinet-layout-and-connection-planning -> 200, 517,177 bytes, full document. Section URLs work; a bare /doc/<EDOC> does not, and no slug for these IDs is published anywhere reachable.

2. The "no OTN port type" blocker recorded in research/optixtrans-board-census.md §2 is wrong, by the catalogue's own precedent. Four merged specs already file a WDM/OTN line side as an Ethernet cage with a prose caveat: ciena/6500-s14 (line1-line4, 200G coherent, ethernet/qsfp28, "modelled at this engine's 100G ceiling"), adva/fsp-3000 (16 coherent line ports), nokia/1830-pss-24x (line1-line8), packetlight/pl-2000m (single 200G uplink). An E9605 (5 x TNG1N210) = 50 x 10G SFP+ line ports can be written in exactly that house style, so no engine change is needed.

Exact missing field: the board->subrack matrix - which board families (TNG1N210 et al.) are valid in the E9605's 5 slots and in the E9612's 13 - i.e. the contents of EDOC1100163901. The partner mirror (ActForNet) reprints it with Y/NA per column and stays a lead, not a source.

What would unblock: anyone with a Huawei support/partner login downloads OptiXtrans E9600 Board Datasheet (EDOC1100163901) and pastes that table, or any indexed section URL of EDOC1100625542 whose slug starts with board. One page converts #807, #808 and #810.

Recorded in research/optixtrans-e9600-board-datasheet.md; PR #1808 carries it. Issue stays open - no spec, no guess.

**The board census now has a vendor URL, a document ID and a 404. And the second blocker on this family was wrong.** **1. Huawei does publish a board datasheet, and the E9605 page links it.** The E9605/E9612/E9624 support pages list, under *Datasheet*, `OptiXtrans E9600 Board Datasheet` -> `/enterprise/en/doc/EDOC1100163901?idPath=24030884|9856748|9858640|250707904|250732995` (`https://support.huawei.com/enterprise/en/optical-transmission/optixtrans-e9605-pid-250732995`, 200, 436,367 bytes, crawler UA, read 2026-11-17). So the matrix is not "unpublished"; it is one document, and that document is withheld from anonymous readers: - `/enterprise/en/doc/EDOC1100163901?idPath=...` -> 302 -> `https://support.huawei.com/view/notFound`, **404**. - `/enterprise/ru/doc/EDOC1100163901?idPath=...` -> **404**. - `/enterprise/en/datasheet/optixtrans-e9600-board-datasheet-pid-250732995` -> 200, but the body is the product page again (slug ignored, `<title>Huawei OptiXtrans E9605 Support Guide...`). - `hedex/hdx.do?docid=EDOC1100163901&lang=en` -> 200, 3,728 bytes, the SPA shell; `&format=json` returns the identical HTML. - Current generation fails the same way: `EDOC1100625542` (V100R026C00 Hardware Description), `EDOC1100622610` (Product Description), `EDOC1100622612` (Feature Description) -> all **404** -> `view/notFound`. Control, so this is not read as "the crawler route broke": `https://support.huawei.com/enterprise/en/doc/EDOC1100318733/4f560405/cabinet-layout-and-connection-planning` -> **200, 517,177 bytes**, full document. Section URLs work; a bare `/doc/<EDOC>` does not, and no slug for these IDs is published anywhere reachable. **2. The "no OTN port type" blocker recorded in `research/optixtrans-board-census.md` §2 is wrong, by the catalogue's own precedent.** Four merged specs already file a WDM/OTN line side as an Ethernet cage with a prose caveat: `ciena/6500-s14` (line1-line4, 200G coherent, `ethernet/qsfp28`, "modelled at this engine's 100G ceiling"), `adva/fsp-3000` (16 coherent line ports), `nokia/1830-pss-24x` (line1-line8), `packetlight/pl-2000m` (single 200G uplink). An `E9605 (5 x TNG1N210)` = 50 x 10G SFP+ line ports can be written in exactly that house style, so no engine change is needed. **Exact missing field:** the board->subrack matrix - which board families (`TNG1N210` et al.) are valid in the E9605's 5 slots and in the E9612's 13 - i.e. the contents of EDOC1100163901. The partner mirror (ActForNet) reprints it with `Y`/`NA` per column and stays a lead, not a source. **What would unblock:** anyone with a Huawei support/partner login downloads OptiXtrans E9600 Board Datasheet (EDOC1100163901) and pastes that table, or any indexed section URL of EDOC1100625542 whose slug starts with `board`. One page converts #807, #808 and #810. Recorded in `research/optixtrans-e9600-board-datasheet.md`; PR #1808 carries it. Issue stays open - no spec, no guess.
Author
Owner

New route found, and the blocker is now one document narrower: the census does not need the partner login, it needs one client-board chapter.

The earlier conclusion in research/optixtrans-e9600-board-datasheet.md was that the board→subrack matrix exists only inside OptiXtrans E9600 Board Datasheet (EDOC1100163901), so a partner account was the only way in. That is wrong. Huawei also publishes the whole OptiXtrans E9600 V100R024C00 Product Documentation as one PDF per board on e-file.huawei.com, anonymously, and every chapter carries a Valid Slots table — which is the frame↔board census this issue has been waiting for.

Read today (vendor-hosted, HTTP 200, application/pdf)

  • https://e-file.huawei.com/marketingcloud/pep/asset/20000001/Material/e75c775ac71144e6b522d7cef1601604/M3T1A669N1188254864652623965/TNG2DAP%20-.pdf — 871,676 B, 51 pp. §1.9 Valid Slots, one occupied slot per board in every frame: OptiXtrans E9605 → IU1, IU2, IU3, IU4, IU5; E9612 → IU1 to IU13; E9624 → IU1 to IU24; E9624X → IU1 to IU24. Same table, footnote: "In the 1:3 mode of an E9624 subrack, service boards cannot be installed in slots IU6, IU7, IU18, and IU19."
  • https://e-file.huawei.com/marketingcloud/pep/asset/20000001/Material/08ed3c9aa1274b2c84d9a591d467aaba/M3T1A669N1178216511632400475/OptiXtrans%20E9600%20TNG2SRAPXF.pdf — 1,104,102 B, 45 pp, same chapter structure.

Two things this settles: the board documentation uses the enterprise subrack names directly (no OSN 9800 M05/M12/M24 mapping has to be assumed), and the board side independently confirms the 1:3 arithmetic — 24 slots minus IU6, IU7, IU18, IU19 is 20, i.e. the brochure's "1:3 cross-connect mode: 10 large slots or 20 small slots".

Exact missing field for #807: the chapter of a client-port board that fits an E9605, i.e. the PDF that carries both its front panel and its Valid Slots row. Search reaches only OA boards (TNG2DAP, TNG2SRAPXF, TNG3SRAPXF) and one transceiver module; queries for TNG1N206, TNG1N210, TNG1N401, TNG1T401, TNG1M210D, TNV1T210U on e.huawei.com / e-file.huawei.com return the family brochure or nothing. An OA board's panel is amplifier IN/OUT — not the 50 client ports this request needs — so no spec is written from what is public today.

Still 404, so nobody re-walks it: EDOC1100163901 on the trailing-slash, zh and idPath shapes (452,479 B not-found page); https://support.huawei.com/carrier/en/doc/EDOC1100163901 answers 200 with a 3,740 B Vue SPA shell and no anonymous content API; EDOC1100625542 (V100R026C00 HD) is 404 as well.

What closes it in one step: any e-file.huawei.com/…/<board>.pdf chapter for a TNG1N*, TNG1T*, TNG1M* or TNV* board — then this becomes OptiXtrans E9605 (5x <board>) with the ports read off that board's own Port Specifications, which is the label shape docs/RESOLUTIONS.md already sets.

Recorded in research/optixtrans-board-chapters-public.md (PR #1816). Issue stays open.

**New route found, and the blocker is now one document narrower: the census does not need the partner login, it needs one client-board chapter.** The earlier conclusion in `research/optixtrans-e9600-board-datasheet.md` was that the board→subrack matrix exists only inside *OptiXtrans E9600 Board Datasheet* (`EDOC1100163901`), so a partner account was the only way in. That is wrong. Huawei also publishes the whole **OptiXtrans E9600 V100R024C00 Product Documentation** as one PDF per board on `e-file.huawei.com`, anonymously, and every chapter carries a **Valid Slots** table — which is the frame↔board census this issue has been waiting for. **Read today (vendor-hosted, HTTP 200, application/pdf)** - `https://e-file.huawei.com/marketingcloud/pep/asset/20000001/Material/e75c775ac71144e6b522d7cef1601604/M3T1A669N1188254864652623965/TNG2DAP%20-.pdf` — 871,676 B, 51 pp. §1.9 Valid Slots, one occupied slot per board in every frame: **OptiXtrans E9605 → IU1, IU2, IU3, IU4, IU5**; E9612 → IU1 to IU13; E9624 → IU1 to IU24; E9624X → IU1 to IU24. Same table, footnote: *"In the 1:3 mode of an E9624 subrack, service boards cannot be installed in slots IU6, IU7, IU18, and IU19."* - `https://e-file.huawei.com/marketingcloud/pep/asset/20000001/Material/08ed3c9aa1274b2c84d9a591d467aaba/M3T1A669N1178216511632400475/OptiXtrans%20E9600%20TNG2SRAPXF.pdf` — 1,104,102 B, 45 pp, same chapter structure. Two things this settles: the board documentation uses the enterprise subrack names directly (no OSN 9800 M05/M12/M24 mapping has to be assumed), and the board side independently confirms the 1:3 arithmetic — 24 slots minus IU6, IU7, IU18, IU19 is 20, i.e. the brochure's "1:3 cross-connect mode: 10 large slots or 20 small slots". **Exact missing field for #807:** the chapter of a **client-port** board that fits an E9605, i.e. the PDF that carries both its front panel and its Valid Slots row. Search reaches only OA boards (`TNG2DAP`, `TNG2SRAPXF`, `TNG3SRAPXF`) and one transceiver module; queries for `TNG1N206`, `TNG1N210`, `TNG1N401`, `TNG1T401`, `TNG1M210D`, `TNV1T210U` on `e.huawei.com` / `e-file.huawei.com` return the family brochure or nothing. An OA board's panel is amplifier IN/OUT — not the 50 client ports this request needs — so no spec is written from what is public today. **Still 404, so nobody re-walks it:** `EDOC1100163901` on the trailing-slash, `zh` and `idPath` shapes (452,479 B not-found page); `https://support.huawei.com/carrier/en/doc/EDOC1100163901` answers 200 with a 3,740 B Vue SPA shell and no anonymous content API; `EDOC1100625542` (V100R026C00 HD) is 404 as well. **What closes it in one step:** any `e-file.huawei.com/…/<board>.pdf` chapter for a `TNG1N*`, `TNG1T*`, `TNG1M*` or `TNV*` board — then this becomes `OptiXtrans E9605 (5x <board>)` with the ports read off that board's own Port Specifications, which is the label shape `docs/RESOLUTIONS.md` already sets. Recorded in `research/optixtrans-board-chapters-public.md` (PR #1816). Issue stays open.
Author
Owner

Re-checked: the search route is now closed by exhaustion, not by bad luck. Still blocked on one file. PR #1843 records this in research/optixtrans-board-chapters-public.md §5; the issue stays open.

What this pass did differently. Instead of searching for board names again, it crawled the graph the earlier hits came from: every https://e.huawei.com/en/documents/others/<32-hex> material page lists its siblings, and breadth-first from the fourteen hashes already on this thread that graph drains in 52 documents. The frontier is empty, so §5.1 is the complete public set, not a sample.

class count what
OptiXtrans frame chapters 4 E9600 family brochure, E9612, E9605, E6608T
OptiXtrans board chapters 3 TNG2DAP, TNG2SRAPXF, TNG3SRAPXF — all OA (amplifier) boards
transceiver datasheets 6 SFP+/eSFP variants
unrelated products on the same hub 38 OceanStor, CloudEngine, IdeaHub, eSpace …

TNG1N206, TNG1N210, TNG1N401, TNG1T*, TNG1M*, TNV*: zero chapters anywhere in it.

TNG3SRAPXF — the one board chapter the previous pass listed as indexed but unfetched — was fetched whole this pass (HTTP 200, application/pdf, 1,228,914 B, 50 pp) in case its Mappings and Substitution table spilled outside the OA class. It does not: §1.3 pairs only TNG2SRAPXF01, TNG3SRAPX, TNG1CXP, TNG3CXP, TNG1HCXP, TNG2HSOCXP, TNG3CXCS02/03 and the TNG3O* OA set. Text hits for TNG1N/TNG1T/TNG1M/TNV: 0/0/0/0.

Two routes closed so nobody re-walks them. e.huawei.com site search (/en/search?q=, /en/search-result?keyword=, /en/searchcenter/search?keyword=, /en/documents?keyword=) answers 200 with the same 336 KB client-rendered shell and 0 occurrences of the query — there is no server-side index to page. info.support.huawei.com hedex paths (/hedex/api/pages|navi|book|menu/<EDOC>) all return the same 91,294 B Vue shell, and Info-Finder Hardware Center is a 5,804 B shell whose bundle sets baseURL:"/info-finder" with an X-Csrf-Token cookie; a browser user-agent gets 403 on both hosts.

A second source checked and rejected. ENISA EUCC-3089-2026-004, Certification Report (https://certification.enisa.europa.eu/document/download/e4a5ea1c-266b-4847-a8da-bf114e8dc6ff_en?filename=EUCC-3089-2026-004-CR.pdf, 544,246 B, 25 pp, text layer extracted) names OptiXtrans E9600 Series (E9605/E9612/E9624/E9624X) but contains 0 occurrences of TNG and 0 of board. It certifies software.

Exact missing field, unchanged: one TNG1N*/TNG1M* chapter's Front Panel + Port Specifications. What would unblock it now: a stated configuration — e.g. "E9605, DC master, 5 x TNG1N210" — which converts to OptiXtrans E9605 (5x TNG1N210) = 50 x 10G SFP+ line ports on the numbers already transcribed in research/optixtrans-subrack-chapters-public.md; or a support/partner login for EDOC1100163901 OptiXtrans E9600 Board Datasheet, the one document that carries the board→subrack matrix. Not another search: this thread has now shown the public set is complete for OA boards and empty for every board class this shelf can carry service on. Same applies to #808 and #810.

**Re-checked: the search route is now closed by exhaustion, not by bad luck. Still blocked on one file.** PR #1843 records this in `research/optixtrans-board-chapters-public.md` §5; the issue stays open. **What this pass did differently.** Instead of searching for board names again, it crawled the graph the earlier hits came from: every `https://e.huawei.com/en/documents/others/<32-hex>` material page lists its siblings, and breadth-first from the fourteen hashes already on this thread that graph **drains in 52 documents**. The frontier is empty, so §5.1 is the complete public set, not a sample. | class | count | what | |--|--|--| | OptiXtrans frame chapters | 4 | E9600 family brochure, E9612, **E9605**, E6608T | | OptiXtrans **board** chapters | 3 | `TNG2DAP`, `TNG2SRAPXF`, `TNG3SRAPXF` — all OA (amplifier) boards | | transceiver datasheets | 6 | SFP+/eSFP variants | | unrelated products on the same hub | 38 | OceanStor, CloudEngine, IdeaHub, eSpace … | `TNG1N206`, `TNG1N210`, `TNG1N401`, `TNG1T*`, `TNG1M*`, `TNV*`: **zero chapters anywhere in it.** `TNG3SRAPXF` — the one board chapter the previous pass listed as indexed but unfetched — was fetched whole this pass (`HTTP 200`, `application/pdf`, **1,228,914 B, 50 pp**) in case its *Mappings and Substitution* table spilled outside the OA class. It does not: §1.3 pairs only `TNG2SRAPXF01`, `TNG3SRAPX`, `TNG1CXP`, `TNG3CXP`, `TNG1HCXP`, `TNG2HSOCXP`, `TNG3CXCS02/03` and the `TNG3O*` OA set. Text hits for `TNG1N`/`TNG1T`/`TNG1M`/`TNV`: 0/0/0/0. **Two routes closed so nobody re-walks them.** `e.huawei.com` site search (`/en/search?q=`, `/en/search-result?keyword=`, `/en/searchcenter/search?keyword=`, `/en/documents?keyword=`) answers 200 with the same 336 KB client-rendered shell and **0** occurrences of the query — there is no server-side index to page. `info.support.huawei.com` `hedex` paths (`/hedex/api/pages|navi|book|menu/<EDOC>`) all return the same **91,294 B** Vue shell, and Info-Finder Hardware Center is a 5,804 B shell whose bundle sets `baseURL:"/info-finder"` with an `X-Csrf-Token` cookie; a browser user-agent gets 403 on both hosts. **A second source checked and rejected.** ENISA EUCC-3089-2026-004, *Certification Report* (`https://certification.enisa.europa.eu/document/download/e4a5ea1c-266b-4847-a8da-bf114e8dc6ff_en?filename=EUCC-3089-2026-004-CR.pdf`, 544,246 B, 25 pp, text layer extracted) names `OptiXtrans E9600 Series (E9605/E9612/E9624/E9624X)` but contains **0** occurrences of `TNG` and **0** of `board`. It certifies software. **Exact missing field, unchanged:** one `TNG1N*`/`TNG1M*` chapter's *Front Panel* + *Port Specifications*. **What would unblock it now:** a stated configuration — e.g. *"E9605, DC master, 5 x TNG1N210"* — which converts to `OptiXtrans E9605 (5x TNG1N210)` = 50 x 10G SFP+ line ports on the numbers already transcribed in `research/optixtrans-subrack-chapters-public.md`; or a support/partner login for `EDOC1100163901` *OptiXtrans E9600 Board Datasheet*, the one document that carries the board→subrack matrix. Not another search: this thread has now shown the public set is complete for OA boards and empty for every board class this shelf can carry service on. Same applies to #808 and #810.
Author
Owner

Still blocked, but the thread's stated blocker was wrong and is now corrected in the queue record — PR #1849. This issue stays open.

What was wrong. This thread said the board-to-subrack matrix (EDOC1100163901) was "withheld, not unpublished". The matrix is published, anonymously, on the vendor file host:

https://e-file.huawei.com/marketingcloud/pep/asset/20000001/Material/e75c775ac71144e6b522d7cef1601604/M3T1A669N1188254864652623965/TNG2DAP%20-.pdf — HTTP 200, application/pdf, 871,676 B, 51 pp. §1.9 "Valid Slots", Table 1: OptiXtrans E9605 | 1 | IU1, IU2, IU3, IU4, IU5 (E9612 IU1 to IU13, E9624 IU1 to IU24), plus "In the 1:3 mode of an E9624 subrack, service boards cannot be installed in slots IU6, IU7, IU18, and IU19."

So the frame half is done: 5 service slots, each holding one board, and E9605 accepts the line boards (TNG1N206, TNG1N210, TNG1N401), transponder units (TNG1M*), amplifiers and mux/ROADM boards.

The field that is still missing: one CLIENT board chapter. Every e-file chapter Huawei indexes is an optical-layer board — TNG2DAP is a 5x LC EDFA, and the asset paths are opaque (<32-hex hash>/M3T1A669N<19 digits>/), so a chapter cannot be guessed from a sibling. Searched today, no hit: TNG1N206, TNG1N210, TNG1N401, TNG1M210D, TNG1M404DM, TNG1M402, TNG1M520SM on e.huawei.com and e-file.huawei.com. The account-walled EDOC1100163901 / EDOC1100621488 remain the only indexes of the board set.

What would unblock it, precisely: that one chapter PDF (its "Port Specifications" table), or a stated configuration plus the board's port table. The board names print their own counts — "TNG1N210: 10 x 10 Gbit/s line service processing board" — so E9605 (5x TNG1N210) becomes writable the moment a vendor port table corroborates the name. Writing it from the partner mirror plus the name convention alone is the family-sibling guess this catalogue has corrected three times, which is why I did not.

#808 and #810 were corrected in the same commit: their frame arithmetic is also vendor-sourced now, and #810 additionally needs the protection mode (1:1 → 12 large/24 small slots, 1:3 → 10/20).

**Still blocked, but the thread's stated blocker was wrong and is now corrected in the queue record — PR #1849. This issue stays open.** **What was wrong.** This thread said the board-to-subrack matrix (`EDOC1100163901`) was "withheld, not unpublished". The matrix is published, anonymously, on the vendor file host: `https://e-file.huawei.com/marketingcloud/pep/asset/20000001/Material/e75c775ac71144e6b522d7cef1601604/M3T1A669N1188254864652623965/TNG2DAP%20-.pdf` — HTTP 200, `application/pdf`, 871,676 B, 51 pp. §1.9 "Valid Slots", Table 1: `OptiXtrans E9605 | 1 | IU1, IU2, IU3, IU4, IU5` (E9612 `IU1 to IU13`, E9624 `IU1 to IU24`), plus "In the 1:3 mode of an E9624 subrack, service boards cannot be installed in slots IU6, IU7, IU18, and IU19." So the **frame half is done**: 5 service slots, each holding one board, and E9605 accepts the line boards (`TNG1N206`, `TNG1N210`, `TNG1N401`), transponder units (`TNG1M*`), amplifiers and mux/ROADM boards. **The field that is still missing: one CLIENT board chapter.** Every e-file chapter Huawei indexes is an optical-layer board — `TNG2DAP` is a 5x LC EDFA, and the asset paths are opaque (`<32-hex hash>/M3T1A669N<19 digits>/`), so a chapter cannot be guessed from a sibling. Searched today, no hit: `TNG1N206`, `TNG1N210`, `TNG1N401`, `TNG1M210D`, `TNG1M404DM`, `TNG1M402`, `TNG1M520SM` on `e.huawei.com` and `e-file.huawei.com`. The account-walled `EDOC1100163901` / `EDOC1100621488` remain the only indexes of the board set. **What would unblock it, precisely:** that one chapter PDF (its "Port Specifications" table), or a stated configuration *plus* the board's port table. The board names print their own counts — "TNG1N210: 10 x 10 Gbit/s line service processing board" — so `E9605 (5x TNG1N210)` becomes writable the moment a vendor port table corroborates the name. Writing it from the partner mirror plus the name convention alone is the family-sibling guess this catalogue has corrected three times, which is why I did not. `#808` and `#810` were corrected in the same commit: their frame arithmetic is also vendor-sourced now, and #810 additionally needs the protection mode (1:1 → 12 large/24 small slots, 1:3 → 10/20).
Author
Owner

Still blocked, and the portal is now closed by counting rather than by sampling. One vendor sentence was gained, and it changes which documents would be acceptable. PR #1884 records this; it does not close the issue.

1. The support portal serves 146 of its 174 documents to nobody. The OSN 9800 M24 page (https://support.huawei.com/enterprise/en/optical-transmission/optix-osn-9800-m24-pid-22048459, 200, 442,970 B) links 174 unique document ids. Requested one at a time, 146 answer 302 -> /view/notFound, and the page's own Datasheet entry — OptiX OSN 9800 Board Datasheet -> EDOC1000074739 — is one of them. So the 2014-era EDOC1000… numbering is as closed as the current EDOC11006…: "wait for the old documents to go public" was the last untested hope in this thread and it is now tested.

2. The 28 that open carry no ports. 24 are videos (product introduction, OSU-advantage set, twelve Optical Basics issues). The two documents are EDOC1100605036 User Guide (550,913 B) and EDOC1100605336 EUCC statement (462,529 B); in the User Guide the strings TNG1N, Port Specifications, IU1 and line service processing occur zero times, and E9605/E9612/E9624 once each — in scope text.

3. What those two documents do give, and its exact limit. Both scope themselves as: "The hardware models covered by this guidance are: The OSN 9800 series (U64E/U32E/M24/M12/M05/P32/P32C) and E9600 series (E9605/E9612/E9624/E9624X)." Vendor-authored, on a Huawei host: the E9600 subracks are the same hardware family as the OSN 9800 M subracks, so a board document filed under OptiX OSN 9800 covers them — the E9624X trap does not apply to it. It prints no E9605 -> M05 pairing, so it is still not permission to take the M05 census.

4. carrier.huawei.com checked and closed. Its whole networks/ PDF folder is 103 archived URLs, 8 WDM/OTN, all brochures: the OSN 9800 M datasheet text layer (14,161 chars, 6 pp) has 0 TNG1N and 0 N210. The live host 403s every User-Agent tried, and the archived brochure capture is truncated at 1 MiB — noted so nobody spends two fetches on it again.

One operational note for the next pass: this portal 403s (472 B) browser and curl User-Agents on product pages while serving a crawler UA 442,970 B. A browser-UA probe here measures the WAF, not the catalogue; three earlier notes record 403s on this family that were User-Agent-dependent.

Exact missing field, unchanged: one TNG1N206 (6 x 10G) / TNG1N210 (10 x 10G) / TNG1N401 (1 x 100G) chapter with a Port Specifications table. Unblock: that chapter (an OSN 9800 board chapter now counts per §3), or EDOC1100163901, pasted by a reader with a support session — or a stated configuration plus the board's port table.

**Still blocked, and the portal is now closed by counting rather than by sampling. One vendor sentence was gained, and it changes which documents would be acceptable. PR #1884 records this; it does not close the issue.** **1. The support portal serves 146 of its 174 documents to nobody.** The OSN 9800 M24 page (`https://support.huawei.com/enterprise/en/optical-transmission/optix-osn-9800-m24-pid-22048459`, 200, 442,970 B) links 174 unique document ids. Requested one at a time, **146 answer 302 -> `/view/notFound`**, and the page's own Datasheet entry — *OptiX OSN 9800 Board Datasheet* -> `EDOC1000074739` — is one of them. So the 2014-era `EDOC1000…` numbering is as closed as the current `EDOC11006…`: "wait for the old documents to go public" was the last untested hope in this thread and it is now tested. **2. The 28 that open carry no ports.** 24 are videos (product introduction, OSU-advantage set, twelve `Optical Basics` issues). The two documents are `EDOC1100605036` *User Guide* (550,913 B) and `EDOC1100605336` *EUCC statement* (462,529 B); in the User Guide the strings `TNG1N`, `Port Specifications`, `IU1` and `line service processing` occur **zero** times, and `E9605`/`E9612`/`E9624` once each — in scope text. **3. What those two documents do give, and its exact limit.** Both scope themselves as: *"The hardware models covered by this guidance are: The OSN 9800 series (U64E/U32E/M24/M12/M05/P32/P32C) and E9600 series (E9605/E9612/E9624/E9624X)."* Vendor-authored, on a Huawei host: the E9600 subracks are the **same hardware family** as the OSN 9800 M subracks, so a board document filed under *OptiX OSN 9800* covers them — the E9624X trap does not apply to it. It prints no `E9605 -> M05` pairing, so it is still not permission to take the M05 census. **4. carrier.huawei.com checked and closed.** Its whole `networks/` PDF folder is 103 archived URLs, 8 WDM/OTN, all brochures: the OSN 9800 M datasheet text layer (14,161 chars, 6 pp) has **0** `TNG1N` and **0** `N210`. The live host 403s every User-Agent tried, and the archived brochure capture is truncated at 1 MiB — noted so nobody spends two fetches on it again. **One operational note for the next pass:** this portal 403s (472 B) browser and curl User-Agents on product pages while serving a crawler UA 442,970 B. A browser-UA probe here measures the WAF, not the catalogue; three earlier notes record 403s on this family that were User-Agent-dependent. **Exact missing field, unchanged:** one `TNG1N206` (6 x 10G) / `TNG1N210` (10 x 10G) / `TNG1N401` (1 x 100G) chapter with a Port Specifications table. **Unblock:** that chapter (an OSN 9800 board chapter now counts per §3), or `EDOC1100163901`, pasted by a reader with a support session — or a stated configuration *plus* the board's port table.
Author
Owner

Still blocked, and the search space is now closed by counting on this product's own page rather than a neighbour's. PR #1887 records it; the issue stays open.

1. The E9605 page links 149 documents; 110 of them are withheld, including the Board Datasheet.
https://support.huawei.com/enterprise/en/optical-transmission/optixtrans-e9605-pid-250732995 — HTTP 200, 436,367 B, crawler user-agent, read 2026-11-17. Requested one at a time:

result count what
200 39 24 (Video), 12 【Optical Basics】 issues, 3 documents
302 -> /view/notFound 110 everything else, incl. EDOC1100163901 OptiXtrans E9600 Board Datasheet

Text census across all 39 public bodies: TNG1 0, Port Specifications 0, IU1 0. The three documents are EDOC1100605036 (OSN 9800 + E9600 User Guide), EDOC1100605336 (EUCC statement) and EDOC1100414468 (Smart Urban Rail solution page).

2. The URL shape is not the reason for the 404s — so nobody needs to retry it. Same host, same UA, same minute: …/en/doc/EDOC1100400005 (a guide linked from that same page) → 200, 492,157 B; an article inside it (…/EDOC1100400005/bec365f7/network-management) → 200, 484,645 B. EDOC1100625542 (Hardware Description) and EDOC1100163901 (Board Datasheet) → 302 → /view/notFound. Publication status, not grammar.

3. A tributary board is the cheaper ask, and TNG1A212 is the name to look for. The line boards in this thread (TNG1N206/210/401) all port on the WDM side, so the only expressible value pair is ethernet/sfp+ — a wrong protocol claim on a real socket, which is why this was left unmodelled. A tributary board ports on the client side and carries 10GE LAN as a first-class service, so ethernet, 10G, sfp+ is then the vendor's own claim. TNG1A212 (12 x 10G any-rate; each port usable as a tributary or line port) fits IU1-IU5 and has 12 ports per partner mirrors only (actfornet.com/solutions/solution/OptiXtrans_E9600_Boards_Overview, thunder-link.com/tng1m210d.html) — a lead, not evidence, and not modelled.

Exact missing artifact: one Huawei-hosted chapter for a tributary or transponder board with a Port Specifications table — TNG1A212, TNG1T212 or TNG1M210D. Neither e-file.huawei.com nor e.huawei.com indexes one today (searched 2026-11-17).

Unblock: that one chapter, EDOC1100163901 pasted by a reader with a support session, or a stated configuration plus the board's port table. With the chapter, this closes as OptiXtrans E9605 (5x TNG1A212) = 60 x 10G client ports plus that board's published line ports, slot count from TNG2DAP §1.9 (IU1-IU5) — and it is the same single file for #808 and #810.

**Still blocked, and the search space is now closed by counting on this product's own page rather than a neighbour's. PR #1887 records it; the issue stays open.** **1. The E9605 page links 149 documents; 110 of them are withheld, including the Board Datasheet.** `https://support.huawei.com/enterprise/en/optical-transmission/optixtrans-e9605-pid-250732995` — HTTP 200, 436,367 B, crawler user-agent, read 2026-11-17. Requested one at a time: | result | count | what | |--|--|--| | `200` | **39** | 24 `(Video)`, 12 `【Optical Basics】` issues, 3 documents | | `302 -> /view/notFound` | **110** | everything else, incl. `EDOC1100163901` *OptiXtrans E9600 Board Datasheet* | Text census across all 39 public bodies: `TNG1` **0**, `Port Specifications` **0**, `IU1` **0**. The three documents are `EDOC1100605036` (OSN 9800 + E9600 User Guide), `EDOC1100605336` (EUCC statement) and `EDOC1100414468` (Smart Urban Rail solution page). **2. The URL shape is not the reason for the 404s** — so nobody needs to retry it. Same host, same UA, same minute: `…/en/doc/EDOC1100400005` (a guide linked from that same page) → **200**, 492,157 B; an article inside it (`…/EDOC1100400005/bec365f7/network-management`) → **200**, 484,645 B. `EDOC1100625542` (Hardware Description) and `EDOC1100163901` (Board Datasheet) → 302 → `/view/notFound`. Publication status, not grammar. **3. A tributary board is the cheaper ask, and `TNG1A212` is the name to look for.** The line boards in this thread (`TNG1N206/210/401`) all port on the **WDM side**, so the only expressible value pair is `ethernet`/`sfp+` — a wrong protocol claim on a real socket, which is why this was left unmodelled. A tributary board ports on the **client** side and carries 10GE LAN as a first-class service, so `ethernet`, 10G, `sfp+` is then the vendor's own claim. `TNG1A212` (12 x 10G any-rate; each port usable as a tributary or line port) fits `IU1-IU5` and has 12 ports **per partner mirrors only** (`actfornet.com/solutions/solution/OptiXtrans_E9600_Boards_Overview`, `thunder-link.com/tng1m210d.html`) — a lead, not evidence, and not modelled. **Exact missing artifact:** one Huawei-hosted chapter for a tributary or transponder board with a Port Specifications table — `TNG1A212`, `TNG1T212` or `TNG1M210D`. Neither `e-file.huawei.com` nor `e.huawei.com` indexes one today (searched 2026-11-17). **Unblock:** that one chapter, `EDOC1100163901` pasted by a reader with a support session, or a stated configuration plus the board's port table. With the chapter, this closes as `OptiXtrans E9605 (5x TNG1A212)` = 60 x 10G client ports plus that board's published line ports, slot count from `TNG2DAP` §1.9 (`IU1-IU5`) — and it is the same single file for #808 and #810.
Author
Owner

Still blocked, and the tool Huawei points at for board specs is now tested rather than assumed. PR #1889 records it; the issue stays open.

1. The zh support locale is a second document list, larger than the English one and no more open. https://support.huawei.com/enterprise/zh/optical-transmission/optixtrans-e9605-pid-250732995 — 200, 430,883 B, crawler UA. It links 136 unique EDOC ids; 58 answer 200 at …/zh/doc/<ID>, versus 39 public on the English page. By the page's own titles, 57 of the 58 are 多媒体 / 【知光漫谈】 / 传送小百科 items; the one document is EDOC1100447312, a public infographic. Chinese docs use their own ids (E9600 V100R026C00 产品文档 = EDOC1100621489, not …1488) and are equally gated — and idPath= does not unlock them, so the "this is not a URL-shape problem" reading now holds on a second id family.

2. That infographic names the tool: 光产品Info-Finder … 对比产品/单板/软件规格差异. So the pass went at Info-Finder.

3. Info-Finder's API answers without a session and then refuses every product. …/info-finder/vue/en/enterprise/index (200, 5,804 B; the bare /info-finder/ 301-loops forever) is a Vue SPA over baseURL /info-finder. POST /speccenter/v2/getSpecBoards accepts {"domain":"0","language":"en","pbiId":…} — and returns 10003 outchain mapping pbiIdMd5 not unique for E9605 250732995, E9612 250733000, E9624 250733004, E9624X 251792361, and a CloudEngine S6730-H24X6C control (250538959), so it is a global upstream fault, not an E9600 gap. domain:"1" returns 10110 用户的权限不足 (login).

Missing artifact, unchanged: one board's port census (TNG1N206, TNG1A212, TNG1T212, TNG1M210D), or EDOC1100163901 / EDOC1100621488 pasted by a reader with a session, or a stated configuration naming the board.

**Still blocked, and the tool Huawei points at for board specs is now tested rather than assumed. PR #1889 records it; the issue stays open.** **1. The zh support locale is a second document list, larger than the English one and no more open.** `https://support.huawei.com/enterprise/zh/optical-transmission/optixtrans-e9605-pid-250732995` — 200, 430,883 B, crawler UA. It links **136** unique `EDOC` ids; **58** answer 200 at `…/zh/doc/<ID>`, versus 39 public on the English page. By the page's own titles, 57 of the 58 are *多媒体* / *【知光漫谈】* / *传送小百科* items; the one document is `EDOC1100447312`, a public infographic. Chinese docs use their own ids (E9600 V100R026C00 产品文档 = `EDOC1100621489`, not `…1488`) and are equally gated — and `idPath=` does not unlock them, so the "this is not a URL-shape problem" reading now holds on a second id family. **2. That infographic names the tool:** *光产品Info-Finder … 对比产品/**单板**/软件规格差异*. So the pass went at Info-Finder. **3. Info-Finder's API answers without a session and then refuses every product.** `…/info-finder/vue/en/enterprise/index` (200, 5,804 B; the bare `/info-finder/` 301-loops forever) is a Vue SPA over `baseURL /info-finder`. `POST /speccenter/v2/getSpecBoards` accepts `{"domain":"0","language":"en","pbiId":…}` — and returns `10003 outchain mapping pbiIdMd5 not unique` for E9605 `250732995`, E9612 `250733000`, E9624 `250733004`, E9624X `251792361`, **and a CloudEngine S6730-H24X6C control** (`250538959`), so it is a global upstream fault, not an E9600 gap. `domain:"1"` returns `10110 用户的权限不足` (login). **Missing artifact, unchanged:** one board's port census (`TNG1N206`, `TNG1A212`, `TNG1T212`, `TNG1M210D`), or `EDOC1100163901` / `EDOC1100621488` pasted by a reader with a session, or a stated configuration naming the board.
Author
Owner

Still blocked, and this pass moved the blocker from "the viewer is opaque" to "the viewer's content API answers, and it answers No doc data found for this document set." PR #1893 records it; it does not close the issue.

1. The hedex content API, and the two ids every earlier probe was missing

The viewer shell (/hedex/hdx.do?..., 200, 3,728 B) loads one bundle, /hedex/lst/js/index-legacy-DaAKRolI.js (200, 361,996 B). It names the content endpoints and their template — /hedex/api/pages/${docId}/${topicLibId}/${topicLibVer}/${topicUrl} (its own RegExp is /hedex/api/pages/[^/]+/([^/]+)/([^/]+)/([^#?]+)), plus /hedex/api/navi/ and /hedex/api/search/. Earlier passes had the document id only; the 302 from any /hedex/pages/... URL hands back the pair in Location (tocLibId=BEJ09047&tocLibVer=03 for the OSN 9800 set).

2. Called with those ids, it refuses these documents

URL result
/hedex/api/pages/EDOC1100126589/BEJ09047/03/resources/help/wdm/9800/hd/line/n210/n210.html 404, 18 B, No doc data found. (same without .html)
/hedex/api/navi/EDOC1100126589/first-level-tree 404, {"status":404,"message":"No doc data found.","data":null}
https://support.huawei.cn/hedex/api/pages/EDOC1100625542/BEJ09047/03/resources/help 401, NOT LOGIN!

The API is not a uniform wall — it separates not served (404) from needs a session (401) — and the .cn mirror repeats both byte-for-byte, so a regional host is not a way round the account. No first-level-tree / topic call will hand a lib/ver pair to a caller with no session, which is exactly what EDOC1100625542 (V100R026C00 Hardware Description) needs.

3. The host that used to leak this is retired. info.support.huawei.com is still indexed as an openly readable hedex content host (e.g. /hedex/api/pages/EDOC1000163559/YEI0812D/19/resources/en-us_topic_0000001238552187.html, and the br/fr E9605 product pages), but every path now answers 302 -> https://support.huawei.com/supportindex/ (200, 91,294 B). Wayback holds no capture of .../hedex/api/pages/EDOC1100625542 or .../EDOC1100163901 (CDX prefix query: 200, 0 bytes), though the host itself is archived from 2021-02-18.

4. This is not a crawling gap. e-file.huawei.com/.../OptiXtrans%20E9600%20TNG2SRAPXF.pdf fetched today: 200, application/pdf, 1,104,102 B, and section 1.9 Valid Slots prints for E9605/E9612/E9624/E9624X. Same family, same server, anonymous access — so a published client-board chapter would be reachable. The TNG1N* chapters are unpublished, not unreachable.

Unchanged: the frame half is vendor-sourced (TNG2DAP section 1.9: E9605 = IU1-IU5; E9605 subrack chapter: 5 service boards, PIU/APIU IU100-IU101, EFI IU71-IU72, fan IU90, 177 x 442 x 260 mm). What would close #807 is one TNG1N*/TNG1M* chapter PDF with a Port Specifications table, or a stated E9605 (5x <board>) from a Huawei document.

**Still blocked, and this pass moved the blocker from "the viewer is opaque" to "the viewer's content API answers, and it answers `No doc data found` for this document set." PR #1893 records it; it does not close the issue.** **1. The hedex content API, and the two ids every earlier probe was missing** The viewer shell (`/hedex/hdx.do?...`, 200, 3,728 B) loads one bundle, `/hedex/lst/js/index-legacy-DaAKRolI.js` (200, 361,996 B). It names the content endpoints and their template — `/hedex/api/pages/${docId}/${topicLibId}/${topicLibVer}/${topicUrl}` (its own RegExp is `/hedex/api/pages/[^/]+/([^/]+)/([^/]+)/([^#?]+)`), plus `/hedex/api/navi/` and `/hedex/api/search/`. Earlier passes had the document id only; the `302` from any `/hedex/pages/...` URL hands back the pair in `Location` (`tocLibId=BEJ09047&tocLibVer=03` for the OSN 9800 set). **2. Called with those ids, it refuses these documents** | URL | result | |---|---| | `/hedex/api/pages/EDOC1100126589/BEJ09047/03/resources/help/wdm/9800/hd/line/n210/n210.html` | **404**, 18 B, `No doc data found.` (same without `.html`) | | `/hedex/api/navi/EDOC1100126589/first-level-tree` | **404**, `{"status":404,"message":"No doc data found.","data":null}` | | `https://support.huawei.cn/hedex/api/pages/EDOC1100625542/BEJ09047/03/resources/help` | **401**, `NOT LOGIN!` | The API is not a uniform wall — it separates *not served* (404) from *needs a session* (401) — and the `.cn` mirror repeats both byte-for-byte, so a regional host is not a way round the account. No `first-level-tree` / `topic` call will hand a lib/ver pair to a caller with no session, which is exactly what `EDOC1100625542` (V100R026C00 Hardware Description) needs. **3. The host that used to leak this is retired.** `info.support.huawei.com` is still indexed as an openly readable hedex content host (e.g. `/hedex/api/pages/EDOC1000163559/YEI0812D/19/resources/en-us_topic_0000001238552187.html`, and the `br`/`fr` E9605 product pages), but every path now answers 302 -> `https://support.huawei.com/supportindex/` (200, 91,294 B). Wayback holds no capture of `.../hedex/api/pages/EDOC1100625542` or `.../EDOC1100163901` (CDX prefix query: 200, 0 bytes), though the host itself is archived from 2021-02-18. **4. This is not a crawling gap.** `e-file.huawei.com/.../OptiXtrans%20E9600%20TNG2SRAPXF.pdf` fetched today: 200, `application/pdf`, 1,104,102 B, and section 1.9 Valid Slots prints for E9605/E9612/E9624/E9624X. Same family, same server, anonymous access — so a published client-board chapter would be reachable. The `TNG1N*` chapters are unpublished, not unreachable. **Unchanged:** the frame half is vendor-sourced (TNG2DAP section 1.9: E9605 = IU1-IU5; E9605 subrack chapter: 5 service boards, PIU/APIU IU100-IU101, EFI IU71-IU72, fan IU90, 177 x 442 x 260 mm). What would close #807 is one `TNG1N*`/`TNG1M*` chapter PDF with a Port Specifications table, or a stated `E9605 (5x <board>)` from a Huawei document.
Author
Owner

Re-checked 2026-11-17: the stated-configuration route does not exist for this family, and that is now measured rather than assumed. Still blocked on one client-board chapter — issue stays open. PR #1903 carries the note (research/optixtrans-stated-config-absent.md); it closes nothing.

1. The NetEngine trick has no OptiXtrans equivalent. devices/huawei/netengine-8000-m14.json copies a build because the M14 datasheet prints part numbers with counts in a power row (2*CR5DE2NE4X14+1*CR5D00E8XM25+3*CR5D00LAXF91+1*CR5D00EAGF95=905.6 W). The OptiXtrans documents use the same phrase and carry no complement:

  • OptiXtrans E9600 Family brochure (https://e-file.huawei.com/marketingcloud/pep/asset/20000001/Material/0f9e00e3f2ca431693884a01667d7ec6/M3T1A669N1188254857740624027/OptiXtrans%20E9600%20Family%20-.pdf, 200, 277,705 B) — "typical product configuration" is footnote c on the MTBF rows; TNG occurs 0 times, TMB 0.
  • The E9605 subrack chapter itself declines a default: "Service boards need to be configured based on the service plan and all of them are installed in the service board area." (.../Material/cc7356212e6146b58f38b74dd1102e08/M3T1A669N1188254859704168616/OptiXtrans%20E9605%20-.pdf, 200, 117,447 B, 2 pp).

2. Every open document on the sibling product pages, counted today. E9612 + E9624 support pages: 164 unique EDOC ids, requested one at a time at /enterprise/en/doc/<ID> → 45 open (30 of them video / 【Optical Basics】), 119 → /view/notFound. E6616X page: 260 ids → 47 open, 213 withheld. A body scan across all 92 open documents for TNG1*, TNG2*, TMB1*, TMK*, TM65*, TNK*, TNV* returns nothing. EDOC1100163901 (Board Datasheet), EDOC1100621488 (V100R026C00 Product Documentation) and EDOC1100126589 all still answer 302 → /view/notFound today.

3. A correction, so the next pass does not trust it. The e.huawei.com sitemaps do not index these material pages at all: all 32 sitemap files (111,538 <loc>, 4,151 unique document pages) were downloaded and all 836 enterprise-optical-network / documents/others pages fetched — none of the five known OptiXtrans material hashes appears. So "not in the public index" is not evidence that a chapter is unpublished, and §5 of research/optixtrans-client-board-chapters.md should not be leaned on.

Exact missing field, unchanged: one client-port board chapter — TNG1N206, TNG1N210, TNG1N401, TNG1A212, TNG1T212 or TNG1M210D — carrying its Front Panel and Port Specifications, or one vendor sentence of the form E9605, DC master, 5 x <board>.

What would unblock it: that chapter PDF on e-file.huawei.com, or EDOC1100163901 / EDOC1100621488 pasted by a reader with a support session.

One build I deliberately did not write. OptiXtrans E9605 (5x TNG2DAP) is expressible from public vendor material (TNG2DAP §1.9 Valid Slots IU1–IU5, Port Specifications IN/OUT/VI/VO/MON, all LC). It is not a stated build — Huawei states the board fits the slots, not that a five-board shelf is a product — and TNG2DAP is an EDFA, so it would answer this request with an amplifier shelf. That is the family-sibling guess the catalogue has already had to correct three times.

**Re-checked 2026-11-17: the stated-configuration route does not exist for this family, and that is now measured rather than assumed. Still blocked on one client-board chapter — issue stays open.** PR #1903 carries the note (`research/optixtrans-stated-config-absent.md`); it closes nothing. **1. The NetEngine trick has no OptiXtrans equivalent.** `devices/huawei/netengine-8000-m14.json` copies a build because the M14 datasheet prints part numbers *with counts* in a power row (`2*CR5DE2NE4X14+1*CR5D00E8XM25+3*CR5D00LAXF91+1*CR5D00EAGF95=905.6 W`). The OptiXtrans documents use the same phrase and carry no complement: - *OptiXtrans E9600 Family* brochure (`https://e-file.huawei.com/marketingcloud/pep/asset/20000001/Material/0f9e00e3f2ca431693884a01667d7ec6/M3T1A669N1188254857740624027/OptiXtrans%20E9600%20Family%20-.pdf`, 200, 277,705 B) — "typical product configuration" is footnote `c` on the MTBF rows; `TNG` occurs 0 times, `TMB` 0. - The E9605 subrack chapter itself declines a default: *"Service boards need to be configured based on the service plan and all of them are installed in the service board area."* (`.../Material/cc7356212e6146b58f38b74dd1102e08/M3T1A669N1188254859704168616/OptiXtrans%20E9605%20-.pdf`, 200, 117,447 B, 2 pp). **2. Every open document on the sibling product pages, counted today.** E9612 + E9624 support pages: 164 unique `EDOC` ids, requested one at a time at `/enterprise/en/doc/<ID>` → **45 open** (30 of them video / *【Optical Basics】*), **119 → `/view/notFound`**. E6616X page: 260 ids → 47 open, 213 withheld. A body scan across all 92 open documents for `TNG1*`, `TNG2*`, `TMB1*`, `TMK*`, `TM65*`, `TNK*`, `TNV*` returns **nothing**. `EDOC1100163901` (Board Datasheet), `EDOC1100621488` (V100R026C00 Product Documentation) and `EDOC1100126589` all still answer `302 → /view/notFound` today. **3. A correction, so the next pass does not trust it.** The e.huawei.com sitemaps do not index these material pages at all: all 32 sitemap files (111,538 `<loc>`, 4,151 unique document pages) were downloaded and all 836 `enterprise-optical-network` / `documents/others` pages fetched — none of the five known OptiXtrans material hashes appears. So "not in the public index" is not evidence that a chapter is unpublished, and §5 of `research/optixtrans-client-board-chapters.md` should not be leaned on. **Exact missing field, unchanged:** one client-port board chapter — `TNG1N206`, `TNG1N210`, `TNG1N401`, `TNG1A212`, `TNG1T212` or `TNG1M210D` — carrying its Front Panel and Port Specifications, or one vendor sentence of the form `E9605, DC master, 5 x <board>`. **What would unblock it:** that chapter PDF on `e-file.huawei.com`, or `EDOC1100163901` / `EDOC1100621488` pasted by a reader with a support session. **One build I deliberately did not write.** `OptiXtrans E9605 (5x TNG2DAP)` is expressible from public vendor material (TNG2DAP §1.9 Valid Slots IU1–IU5, Port Specifications IN/OUT/VI/VO/MON, all LC). It is not a stated build — Huawei states the *board* fits the slots, not that a five-board shelf is a product — and TNG2DAP is an EDFA, so it would answer this request with an amplifier shelf. That is the family-sibling guess the catalogue has already had to correct three times.
Author
Owner

Correction to the 2026-11-17 comment on this thread that named a port type as the missing field. That field is not missing and never was a blocker — the catalogue already files OTN/WDM line ports the way the engine requires, and has done since it had any optical transport at all. Recorded in research/otn-line-ports-need-no-new-type.md (PR #1938), which also carries a dated correction block in research/optixtrans-board-census.md section 2.

Four specs on main (efb40710), measured today:

spec line ports filed as
devices/huawei/optixtrans-e6616.json 28 line* ethernet / cfp2 / 200G
devices/ciena/waveserver-ai.json 6 line* ethernet / qsfp28 / 100G
devices/adva/fsp-3000.json 16 line* ethernet / qsfp28 / 100G
devices/packetlight/pl-2000m.json 1 line* ethernet / qsfp28 / 100G

None of those carries Ethernet — a coherent wavelength on an FSP 3000 line card is an OTUk signal — and all four pass scripts/audit-port-types.js and test/hardware-invariants.test.js. The first row is this family's own sibling, and its notes states the compromise explicitly. So the engine's lack of an otn type is a real limitation worth an engine PR (it would let the validator refuse a WDM port joined to a 10GBASE-LR peer), but it is an improvement waiting for that PR, not a prerequisite here.

What this leaves. The blocker on this thread is the one it was before that comment and after it: a TNG1N* / TNG1M* chapter that states the board's front panel and which subrack accepts it, or one stated build. The label that build produces is already fixed by the frame chapter — OptiXtrans E9605 (5x TNG1N210) = 50 x 10G line ports — so nothing further has to be designed once the census is readable. No status change here; the request stays open.

**Correction to the 2026-11-17 comment on this thread that named a *port type* as the missing field.** That field is not missing and never was a blocker — the catalogue already files OTN/WDM line ports the way the engine requires, and has done since it had any optical transport at all. Recorded in `research/otn-line-ports-need-no-new-type.md` (PR #1938), which also carries a dated correction block in `research/optixtrans-board-census.md` section 2. Four specs on `main` (`efb40710`), measured today: | spec | line ports | filed as | |--|--|--| | `devices/huawei/optixtrans-e6616.json` | 28 `line*` | `ethernet` / `cfp2` / 200G | | `devices/ciena/waveserver-ai.json` | 6 `line*` | `ethernet` / `qsfp28` / 100G | | `devices/adva/fsp-3000.json` | 16 `line*` | `ethernet` / `qsfp28` / 100G | | `devices/packetlight/pl-2000m.json` | 1 `line*` | `ethernet` / `qsfp28` / 100G | None of those carries Ethernet — a coherent wavelength on an FSP 3000 line card is an OTUk signal — and all four pass `scripts/audit-port-types.js` and `test/hardware-invariants.test.js`. The first row is this family's own sibling, and its `notes` states the compromise explicitly. So the engine's lack of an `otn` type is a real limitation worth an engine PR (it would let the validator refuse a WDM port joined to a 10GBASE-LR peer), but it is an improvement waiting for that PR, not a prerequisite here. **What this leaves.** The blocker on this thread is the one it was before that comment and after it: a `TNG1N*` / `TNG1M*` chapter that states the board's front panel *and* which subrack accepts it, or one stated build. The label that build produces is already fixed by the frame chapter — `OptiXtrans E9605 (5x TNG1N210)` = 50 x 10G line ports — so nothing further has to be designed once the census is readable. No status change here; the request stays open.
Author
Owner

Built, not blocked. PR #1955 models the E9605 as the stated build 5x TNG1N210 - 50x 10G line ports - and its body starts with Fixes #807, so merging it closes this request.

The two fields earlier passes held apart turn out to be one reading. The OptiXtrans E9600 (E9624/E9612/E9605) Boards Overview chapter (EDOC1100126589BEJ09047) prints, in Table 3, TNG1N206 TNG1N210 | 1:1 Y 1:3 Y | Y | Y - valid in the E9605 subrack - and in Table 1 names the board TNG1N210: 10 x 10 Gbit/s Line Service Processing Board. Five service slots (vendor product page, 177 mm subrack, cross-connect NA) x ten ports = 50. TNG2DAP section 1.9 supplies only the IU1-IU5 numbering, never validity.

Read from the verbatim partner copy at https://www.actfornet.com/solutions/solution/OptiXtrans_E9600_Boards_Overview (226,637 B, 2026-11-18) because the vendor copy EDOC1100163901 still answers 302 -> /view/notFound; the spec notes say exactly which fields are mirror-read and which are huawei.com-read. A readable vendor copy would replace the mirror-read rows, and it would also settle #808 and #810 - those two still wait on their build (E9612 13 slots only in the E3CTU DC chassis; E9624 20 boards in 1:3 mode), not on the board census.

**Built, not blocked.** PR #1955 models the E9605 as the stated build `5x TNG1N210` - 50x 10G line ports - and its body starts with `Fixes #807`, so merging it closes this request. The two fields earlier passes held apart turn out to be one reading. The OptiXtrans E9600 (E9624/E9612/E9605) Boards Overview chapter (EDOC1100126589BEJ09047) prints, in Table 3, `TNG1N206 TNG1N210 | 1:1 Y 1:3 Y | Y | Y` - valid in the E9605 subrack - and in Table 1 names the board `TNG1N210: 10 x 10 Gbit/s Line Service Processing Board`. Five service slots (vendor product page, 177 mm subrack, cross-connect NA) x ten ports = 50. `TNG2DAP` section 1.9 supplies only the IU1-IU5 numbering, never validity. Read from the verbatim partner copy at https://www.actfornet.com/solutions/solution/OptiXtrans_E9600_Boards_Overview (226,637 B, 2026-11-18) because the vendor copy `EDOC1100163901` still answers 302 -> /view/notFound; the spec notes say exactly which fields are mirror-read and which are huawei.com-read. A readable vendor copy would replace the mirror-read rows, and it would also settle #808 and #810 - those two still wait on their build (E9612 13 slots only in the E3CTU DC chassis; E9624 20 boards in 1:3 mode), not on the board census.
Sign in to join this conversation.
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#807
No description provided.