research(optixtrans): the vendor sitemap counts 713 documents and holds two board chapters, both amplifiers #1855

Merged
jaylouisw merged 1 commit from catalogue/1117-1355 into main 2026-11-17 14:28:39 +00:00
Owner

Refs #803, #807, #808, #810 - documents why the family is still blocked; no spec is written and no issue is closed by this PR.

What changed

Four passes looked for the E9600 client-board chapter with search engines and each reported "not indexed on e.huawei.com / e-file.huawei.com". That is a statement about a search, not about the vendor's index. This pass counted the index:

  • https://e.huawei.com/robots.txt lists sitemap-ebg-x-default.xml (1,818,143 B, 11,956 <loc>), of which 881 are /en/documents/others/<32-hex> document pages (713 unique).
  • Each is server-rendered and prints its title in an <h1>; all 713 were fetched. 708 carry a title, 5 are empty shells with no attached file.
  • Board chapters in the whole public index: two, TNG2DAP (EDFA) and TNG3SRAPXF (Raman + EDFA). TNG1N206, TNG1N210, TNG1N401, TNG1M*, TNG1T*, TNV*, TNK*, TMB1* occur zero times.

So the client board chapter is absent from a set that has been counted, not merely unfound.

Two corrections to the queue record

The E9612's slot count is not mode-independent - research/open-device-requests.json and the issue threads said it was. OptiXtrans E9612 - (vendor PDF) gives:

layout DC power AC / HVDC power
AUX+SCC IU1-IU12 = 12 IU3-IU12 = 10 (master), 11 (slave)
E3CTU IU1-IU13 = 13 IU3-IU13 = 11

In the case of AC or high-voltage DC power supplies, one APIU board occupies physical service board slots IU1 and IU2, and the occupied logical slot is IU1.

IU1-IU13 in TNG2DAP section 1.9 is the E3CTU DC figure only - the same off-by-two the E6616 AC chassis was already found to have.

The E9605 does have management interfaces. Not in the spec table, but the subrack chapter puts EFI boards in IU71-IU72 and states "The EFI board offers maintenance and management interfaces". The port count is still unpublished, so this unblocks nothing; the request still needs the client board.

Routes closed, so they are not re-walked

  • Asset bucket will not list: /Material/<hash>/, ?list-type=2, ?prefix= -> 403 AccessDenied.
  • info.support.huawei.com/hedex/api/pages/<EDOC>/... answers 200 with the 91,294-byte support index, not the topic.
  • The per-page "related documents" carousel is a topical mix (it lists an OceanStor 5310 sheet and a DAG1000 ATA beside TNG2DAP), not a family chapter set.
  • The vendor file host serves a PDF to an empty User-Agent and refuses a browser one (403, 533 B); python urllib is refused where curl succeeds a second later. That is a client fingerprint, which is why earlier passes recorded the file host as WAF-blocked.

Gates

  • node scripts/syntax-check.js - 388 files, all OK
  • node scripts/audit-port-types.js - no port contradicts its declared type
  • node scripts/check-control-bytes.js - clean
  • node scripts/regen-registry-doc.js --check - pages current (2,348 models / 258 vendors / 29,235 sockets)
  • node scripts/run-tests.js --concurrency=8 - 316 suites, 11,005 assertions; the four canonical-urls failures are that suite's dirty-tree probe firing on this branch's uncommitted files and pass 52/52 once committed

docs/DEVICE_REGISTRY.md is regenerated from research/open-device-requests.json, not hand-edited.

Refs #803, #807, #808, #810 - documents why the family is still blocked; no spec is written and no issue is closed by this PR. ## What changed Four passes looked for the E9600 client-board chapter with search engines and each reported "not indexed on e.huawei.com / e-file.huawei.com". That is a statement about a search, not about the vendor's index. This pass counted the index: * `https://e.huawei.com/robots.txt` lists `sitemap-ebg-x-default.xml` (1,818,143 B, 11,956 `<loc>`), of which **881** are `/en/documents/others/<32-hex>` document pages (**713** unique). * Each is server-rendered and prints its title in an `<h1>`; all 713 were fetched. 708 carry a title, 5 are empty shells with no attached file. * **Board chapters in the whole public index: two**, `TNG2DAP` (EDFA) and `TNG3SRAPXF` (Raman + EDFA). `TNG1N206`, `TNG1N210`, `TNG1N401`, `TNG1M*`, `TNG1T*`, `TNV*`, `TNK*`, `TMB1*` occur **zero** times. So the client board chapter is absent from a set that has been counted, not merely unfound. ## Two corrections to the queue record **The E9612's slot count is not mode-independent** - `research/open-device-requests.json` and the issue threads said it was. `OptiXtrans E9612 -` (vendor PDF) gives: | layout | DC power | AC / HVDC power | |---|---|---| | AUX+SCC | IU1-IU12 = 12 | IU3-IU12 = 10 (master), 11 (slave) | | E3CTU | IU1-IU13 = 13 | IU3-IU13 = 11 | > In the case of AC or high-voltage DC power supplies, one APIU board occupies physical service board slots IU1 and IU2, and the occupied logical slot is IU1. `IU1-IU13` in `TNG2DAP` section 1.9 is the **E3CTU DC** figure only - the same off-by-two the E6616 AC chassis was already found to have. **The E9605 does have management interfaces.** Not in the spec table, but the subrack chapter puts EFI boards in IU71-IU72 and states "The EFI board offers maintenance and management interfaces". The port *count* is still unpublished, so this unblocks nothing; the request still needs the client board. ## Routes closed, so they are not re-walked * Asset bucket will not list: `/Material/<hash>/`, `?list-type=2`, `?prefix=` -> 403 `AccessDenied`. * `info.support.huawei.com/hedex/api/pages/<EDOC>/...` answers 200 with the 91,294-byte support index, not the topic. * The per-page "related documents" carousel is a topical mix (it lists an OceanStor 5310 sheet and a DAG1000 ATA beside TNG2DAP), not a family chapter set. * The vendor file host serves a PDF to an **empty** User-Agent and refuses a browser one (403, 533 B); `python urllib` is refused where `curl` succeeds a second later. That is a client fingerprint, which is why earlier passes recorded the file host as WAF-blocked. ## Gates * `node scripts/syntax-check.js` - 388 files, all OK * `node scripts/audit-port-types.js` - no port contradicts its declared type * `node scripts/check-control-bytes.js` - clean * `node scripts/regen-registry-doc.js --check` - pages current (2,348 models / 258 vendors / 29,235 sockets) * `node scripts/run-tests.js --concurrency=8` - 316 suites, 11,005 assertions; the four `canonical-urls` failures are that suite's dirty-tree probe firing on this branch's uncommitted files and pass 52/52 once committed `docs/DEVICE_REGISTRY.md` is regenerated from `research/open-device-requests.json`, not hand-edited.
research(optixtrans): the vendor sitemap counts 713 documents and holds two board chapters, both amplifiers (#808)
All checks were successful
CI / Classify the change (pull_request) Successful in 8s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 13s
CI / Server smoke test (pull_request) Successful in 7s
CI / Docker build (pull_request) Successful in 23s
CI / Test (Node 20) (pull_request) Successful in 49s
CI / Test (Node 22) (pull_request) Successful in 53s
CI / Test (Node 18) (pull_request) Successful in 57s
CI / Arm auto-merge (pull_request) Successful in 8s
b78b535f52
Four passes looked for the E9600 client-board chapter with search engines and each reported
"not indexed". This pass counted the index instead: e.huawei.com's own sitemap lists 881
/en/documents/others pages (713 unique), every one is server-rendered and carries its title in an
<h1>, and all 713 were fetched. Two board chapters exist in the whole set - TNG2DAP (EDFA) and
TNG3SRAPXF (Raman + EDFA) - and TNG1N*/TNG1M*/TNG1T*/TNV*/TNK*/TMB1* occur zero times. "Absent
from a set that has been counted" is a different fact from "did not turn up in a search", and it
closes the enumeration route the queue has been spending passes on.

Two corrections fall out of the chassis chapters the same enumeration turned up:

  * The E9612's service-slot count is NOT mode-independent, which the queue record said. E3CTU DC
    is IU1-IU13; the older AUX+SCC layout is 12 on DC; and under AC/HVDC power one APIU occupies
    IU1 and IU2 physically, so an E3CTU AC build is IU3-IU13 = 11 and an AUX+SCC build is 10
    (master) / 11 (slave). `IU1-IU13` in TNG2DAP section 1.9 is the E3CTU DC figure only.
  * The E9605 does have management interfaces after all - not in the spec table, but the subrack
    chapter puts EFI boards in IU71-IU72 and states "The EFI board offers maintenance and
    management interfaces". The port count is still unpublished, so nothing is unblocked by it.

Also recorded, so they are not re-walked: the asset bucket refuses to list (S3 AccessDenied), the
hedex static-topic shape returns the 91,294-byte support index for optical EDOCs, the per-page
"related documents" carousel is a topical mix rather than a chapter set, and the vendor file host
serves a PDF to an empty User-Agent while refusing a browser one - a client fingerprint, not a
rate limit, which is why earlier passes read it as a WAF block.

No spec is written: the client board's port table is still unpublished. Refs #803 #807 #808 #810.

Co-authored-by: Hermes Agent <jay@jayw.uk>
jaylouisw force-pushed catalogue/1117-1355 from b78b535f52
All checks were successful
CI / Classify the change (pull_request) Successful in 8s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 13s
CI / Server smoke test (pull_request) Successful in 7s
CI / Docker build (pull_request) Successful in 23s
CI / Test (Node 20) (pull_request) Successful in 49s
CI / Test (Node 22) (pull_request) Successful in 53s
CI / Test (Node 18) (pull_request) Successful in 57s
CI / Arm auto-merge (pull_request) Successful in 8s
to 8e69e5dcab
All checks were successful
CI / Classify the change (pull_request) Successful in 11s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 7s
CI / Server smoke test (pull_request) Successful in 8s
CI / Docker build (pull_request) Successful in 22s
CI / Test (Node 20) (pull_request) Successful in 51s
CI / Test (Node 18) (pull_request) Successful in 55s
CI / Test (Node 22) (pull_request) Successful in 53s
CI / Arm auto-merge (pull_request) Successful in 12s
2026-11-17 14:26:41 +00:00
Compare
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!1855
No description provided.