research(oceanprotect): both E8000 enumeration routes are account-gated, and both are now measured #1873

Merged
jaylouisw merged 1 commit from catalogue/1117-1655 into main 2026-11-17 17:07:56 +00:00
Owner

Refs #742 - this PR records measurements; it does not close the issue.

Issue #742 (Huawei OceanProtect E8000) has been parked on one field: the interface-module slots per controller and the ports on each module type. research/oceanprotect-x-series-port-census.md ranked two routes as still untried - Spare Parts Query, and one rendered section of the E8000 book. Both were measured this pass.

1. The document tree needs a Huawei account, at the service

GET https://support.huawei.com/supportgateway/view/v1/enterprise/doc/catalogue?nid=EDOC1100486186 -> 401

{"status":401,"code":"support-view-020003","message":"need login."}
x-login-url: https://uniportal.huawei.com/uniportal1/

Crawler UA gets that JSON; browser UA gets Akamai 403. So 403 is the edge and 401 is the document service, i.e. the endpoint is reachable and wants a session. The body is gzipped whatever Accept-Encoding says, so it reads as an empty 91-byte response without curl --compressed - which is how a previous pass could record "no answer" here.

Consequence: there is no anonymous enumeration entry point for revision-scoped topic hashes. A book renders only when a hash is already indexed by a search engine - which is why the X3000 V200R001 Product Description (EDOC1100368234/8587c2ff/overview) renders and EDOC1100486186 does not.

2. Spare Parts Query: public CSRF token, private data

https://info.support.huawei.com/storage/spareparts/ is public (200, bundle main-ZCQDDWQH.js, 2,605,430 B) and names its call: {API_HOST}/storage-gw/rest/itcms/getDataList?appId=spare&dataSetId=SpareParts_Product_Output&nodeId=SpareParts_Product_Output&lang=en&child=true.

request result
GET /storageinfogw/csrf-gw/rest/csrf/getCsrfToken 200, token issued
GET .../getDataList?appId=spare... with Csrf-Token / X-Csrf-Token / csrf-token 401, 0 bytes
same without session cookie / without Referer / with lang: en 401, 0 bytes
GET /storageinfo-gw-v2/csrf-gw/rest/csrf/getCsrfToken 401, no token

Specifications Query (the route recorded for #743, main-H3SH7DII.js, 2,775,994 B) uses the same API_HOST / API_HOST_V2 pair, so one measurement covers both tools.

3. The E8000 data sheet, re-fetched: still no census

scripts/huawei-material-fetch.js resolved the V200R001C20 data sheet from the material page: 2,407,538 B, 4 slides. Slide 3 carries 2 controllers per node, up to 128 nodes, 2 PiB/node, port types 16/32/64 Gbit/s Fibre Channel; 10/25/40/100GE, enclosure dimensions. Keyword counts over the whole deck: slot 0, interface module 0, IOM 0, port 1.

Also recorded so nobody borrows it: the 820 mm controller-enclosure depth coincides with the X6000/X8000 2 U 25-disk enclosure, but equal depth is not a shared backplane - the one document naming all three (EDOC1100305200, Parts Replacement) only makes the 2 U fan module and smart disk enclosure shared FRUs. Six IOM slots imported on that basis would be the Dorado-Table-6-1 error again.

What would close #742

One rendered section of EDOC1100486186 (Product Description), EDOC1100485347 (Quick Installation Guide, dated 2026-09-30 on Huawei's own product page) or EDOC1100486185 (Installation Guide) pasted by a reader with a support session; or a stated configuration such as "E8000 node, 2 controllers, 2 x 4-port 25GE IOM per controller".

Gates

  • node scripts/syntax-check.js - 391 files, all OK
  • node scripts/audit-port-types.js - no port contradicts its declared type
  • node scripts/check-control-bytes.js - clean
  • node test/hardware-invariants.test.js - 26 passed, 0 failed
  • node test/request-queue.test.js - 23 passed, 0 failed
  • node scripts/check-doc-citations.js - exit 0
  • npm run regen:registry-doc - registry section rewritten from research/open-device-requests.json
  • importer tripwire not exercised: no devices/ spec changed in this PR
Refs #742 - this PR records measurements; it does **not** close the issue. Issue #742 (Huawei OceanProtect E8000) has been parked on one field: the interface-module slots per controller and the ports on each module type. `research/oceanprotect-x-series-port-census.md` ranked two routes as still untried - Spare Parts Query, and one rendered section of the E8000 book. Both were measured this pass. ## 1. The document tree needs a Huawei account, at the service `GET https://support.huawei.com/supportgateway/view/v1/enterprise/doc/catalogue?nid=EDOC1100486186` -> **401** ```json {"status":401,"code":"support-view-020003","message":"need login."} x-login-url: https://uniportal.huawei.com/uniportal1/ ``` Crawler UA gets that JSON; browser UA gets Akamai 403. So 403 is the edge and 401 is the document service, i.e. the endpoint is reachable and wants a session. The body is gzipped whatever `Accept-Encoding` says, so it reads as an empty 91-byte response without `curl --compressed` - which is how a previous pass could record "no answer" here. Consequence: there is no anonymous enumeration entry point for revision-scoped topic hashes. A book renders only when a hash is already indexed by a search engine - which is why the X3000 V200R001 Product Description (`EDOC1100368234/8587c2ff/overview`) renders and `EDOC1100486186` does not. ## 2. Spare Parts Query: public CSRF token, private data `https://info.support.huawei.com/storage/spareparts/` is public (200, bundle `main-ZCQDDWQH.js`, 2,605,430 B) and names its call: `{API_HOST}/storage-gw/rest/itcms/getDataList?appId=spare&dataSetId=SpareParts_Product_Output&nodeId=SpareParts_Product_Output&lang=en&child=true`. | request | result | |---|---| | `GET /storageinfogw/csrf-gw/rest/csrf/getCsrfToken` | **200**, token issued | | `GET .../getDataList?appId=spare...` with `Csrf-Token` / `X-Csrf-Token` / `csrf-token` | **401, 0 bytes** | | same without session cookie / without Referer / with `lang: en` | **401, 0 bytes** | | `GET /storageinfo-gw-v2/csrf-gw/rest/csrf/getCsrfToken` | **401**, no token | Specifications Query (the route recorded for #743, `main-H3SH7DII.js`, 2,775,994 B) uses the same `API_HOST` / `API_HOST_V2` pair, so one measurement covers both tools. ## 3. The E8000 data sheet, re-fetched: still no census `scripts/huawei-material-fetch.js` resolved the V200R001C20 data sheet from the material page: **2,407,538 B, 4 slides**. Slide 3 carries 2 controllers per node, up to 128 nodes, 2 PiB/node, port *types* `16/32/64 Gbit/s Fibre Channel; 10/25/40/100GE`, enclosure dimensions. Keyword counts over the whole deck: `slot` **0**, `interface module` **0**, `IOM` **0**, `port` **1**. Also recorded so nobody borrows it: the 820 mm controller-enclosure depth coincides with the X6000/X8000 2 U 25-disk enclosure, but equal depth is not a shared backplane - the one document naming all three (EDOC1100305200, Parts Replacement) only makes the 2 U fan module and smart disk enclosure shared FRUs. Six IOM slots imported on that basis would be the Dorado-Table-6-1 error again. ## What would close #742 One rendered section of `EDOC1100486186` (Product Description), `EDOC1100485347` (Quick Installation Guide, dated 2026-09-30 on Huawei's own product page) or `EDOC1100486185` (Installation Guide) pasted by a reader with a support session; or a stated configuration such as "E8000 node, 2 controllers, 2 x 4-port 25GE IOM per controller". ## Gates - `node scripts/syntax-check.js` - 391 files, all OK - `node scripts/audit-port-types.js` - no port contradicts its declared type - `node scripts/check-control-bytes.js` - clean - `node test/hardware-invariants.test.js` - 26 passed, 0 failed - `node test/request-queue.test.js` - 23 passed, 0 failed - `node scripts/check-doc-citations.js` - exit 0 - `npm run regen:registry-doc` - registry section rewritten from `research/open-device-requests.json` - importer tripwire not exercised: no `devices/` spec changed in this PR
research(oceanprotect): both E8000 enumeration routes are account-gated, and both are now measured (#742)
All checks were successful
CI / Classify the change (pull_request) Successful in 10s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 9s
CI / Server smoke test (pull_request) Successful in 9s
CI / Docker build (pull_request) Successful in 26s
CI / Test (Node 20) (pull_request) Successful in 49s
CI / Test (Node 18) (pull_request) Successful in 1m0s
CI / Test (Node 22) (pull_request) Successful in 57s
CI / Arm auto-merge (pull_request) Successful in 23s
91e32bc270
The E8000 thread has been parked on one field - the interface-module slots per
controller and the ports on each module type - with two untried routes ranked in
research/oceanprotect-x-series-port-census.md: Spare Parts Query, and one rendered
section of the E8000 book. This pass measured both instead of re-ranking them.

- The document tree's catalogue endpoint answers 401 support-view-020003
  "need login." (x-login-url: uniportal) to a crawler UA and Akamai 403 to a
  browser UA. So there is no anonymous enumeration entry point for topic hashes,
  which is why a hash that search engines already indexed renders and EDOC1100486186
  does not.
- Spare Parts Query hands out a CSRF token anonymously (200) and then answers 401
  with an empty body on its own data call, for every header and cookie variant; the
  v2 gateway refuses even the token. Specifications Query (#743) shares the gateway.
- The E8000 V200R001C20 data sheet was re-fetched through scripts/huawei-material-fetch.js
  (2,407,538 B, 4 slides) and still carries `slot` 0, `interface module` 0, `port` 1.

No spec is written: no document read states the E8000's IOM population, and six
slots borrowed from the X6000/X8000 enclosure would be the family-sibling error.
The registry's open-request row for #742 now carries the measured blocker and the
three E8000 document IDs.

Co-authored-by: Hermes Agent <jay@jayw.uk>
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!1873
No description provided.