huawei: model the OptiXtrans E9612 as a stated 13x TNG1N210 build (#808) #1962

Merged
jaylouisw merged 2 commits from catalogue/1118-0856 into main 2026-11-18 09:39:35 +00:00
Owner

Fixes #808

Device: Huawei OptiXtrans E9612 — devices/huawei/optixtrans-e9612.json
Build modelled: 13x TNG1N210 = 130x 10G line ports (stated complement, same convention as E9605 #1955, E6608, E6616).

Census, field by field

field value source
dimensions / slots / cross-connect / line rate / power 442x295x347.2 mm, 13 service slots, Cross-connect Type NA, 10G-1.2T, -48/-60 V DC or 110/220 V AC or 240/336 V HVDC https://e.huawei.com/en/products/optical-transmission/optixtrans-e9600 (200, 344,531 B, E9612 column)
board fits the E9612 TNG1N206/TNG1N210: Y (E9624 is mode-dependent Y/Y; E9612 takes no mode qualifier) E9600 Boards Overview, Table 3
board port count TNG1N210: 10 x 10 Gbit/s Line Service Processing Board same chapter, Table 1
slot map IU1-IU13 TNG2DAP section 1.9, Valid slots, E9612 row (IU numbering only)

13 x 10 = 130. No guessed port, rate, cage or part number.

Why a partner copy is cited, and what it is used for. The chapter is EDOC1100126589BEJ09047; its vendor copy (E9600 Board Datasheet EDOC1100163901) answers 302 -> /view/notFound for every user-agent tried, as do EDOC1100214449, EDOC1100621488, EDOC1100622610, EDOC1100622612. The Table 1/3 rows were read from the verbatim reproduction at https://www.actfornet.com/solutions/solution/OptiXtrans_E9600_Boards_Overview (200, 226,637 B, 127 hedex links). The spec notes say which fields are mirror-read and which are vendor-page-read; the reseller reprint of the N210 chapter (https://www.thunder-link.com/tng1n210.html) agrees on 10 ports, all SFP+ -10G, M12 IU1-IU13.

The 13 is conditional, and the label says it. Subrack chapter (2026-11-17): DC 13 (IU1-IU13), DC master 12, AC/HVDC 10 or 11 — one APIU occupies physical IU1+IU2. So this entry is the DC 13-slot layout; an AC shelf is 20-30 line ports smaller. Recorded in notes, not only in the thread.

Also in this PR

  • research/optixtrans-e9612-stated-build.md — the reading, the Table 3 rows, the gated ids.
  • research/open-device-requests.json — #808 removed; #810 narrowed: validity is now stated (Table 3 Y in both modes), so only the cross-connect mode is missing (1:1 -> 240 ports, 1:3 -> 200).
  • Generated pages regenerated (regen:registry-doc): 2,355 models / 258 vendors / 29,481 sockets / 971 shapes.

Gates: syntax-check 411 files OK · audit-port-types no contradiction · hardware-invariants 26/26 · import-devices-capabilities 34/34 · catalogue-hygiene 25/25 · faceplate-catalogue 14/14 · fuzz-importer 14/14 · importer-roundtrip 26/26 · doc-catalog-counts 43/43 · check-control-bytes clean · regen-registry-doc --check current.

Fixes #808 **Device:** Huawei OptiXtrans E9612 — `devices/huawei/optixtrans-e9612.json` **Build modelled:** 13x TNG1N210 = 130x 10G line ports (stated complement, same convention as E9605 #1955, E6608, E6616). **Census, field by field** | field | value | source | |--|--|--| | dimensions / slots / cross-connect / line rate / power | 442x295x347.2 mm, **13** service slots, **Cross-connect Type NA**, 10G-1.2T, -48/-60 V DC or 110/220 V AC or 240/336 V HVDC | https://e.huawei.com/en/products/optical-transmission/optixtrans-e9600 (200, 344,531 B, E9612 column) | | board fits the E9612 | `TNG1N206`/`TNG1N210`: **Y** (E9624 is mode-dependent Y/Y; E9612 takes no mode qualifier) | E9600 Boards Overview, Table 3 | | board port count | `TNG1N210: 10 x 10 Gbit/s Line Service Processing Board` | same chapter, Table 1 | | slot map | IU1-IU13 | TNG2DAP section 1.9, Valid slots, E9612 row (IU numbering only) | 13 x 10 = 130. No guessed port, rate, cage or part number. **Why a partner copy is cited, and what it is used for.** The chapter is `EDOC1100126589BEJ09047`; its vendor copy (E9600 Board Datasheet `EDOC1100163901`) answers `302 -> /view/notFound` for every user-agent tried, as do `EDOC1100214449`, `EDOC1100621488`, `EDOC1100622610`, `EDOC1100622612`. The Table 1/3 rows were read from the verbatim reproduction at https://www.actfornet.com/solutions/solution/OptiXtrans_E9600_Boards_Overview (200, 226,637 B, 127 hedex links). The spec notes say which fields are mirror-read and which are vendor-page-read; the reseller reprint of the N210 chapter (https://www.thunder-link.com/tng1n210.html) agrees on 10 ports, all SFP+ -10G, M12 IU1-IU13. **The 13 is conditional, and the label says it.** Subrack chapter (2026-11-17): DC 13 (IU1-IU13), DC master 12, AC/HVDC 10 or 11 — one APIU occupies physical IU1+IU2. So this entry is the DC 13-slot layout; an AC shelf is 20-30 line ports smaller. Recorded in `notes`, not only in the thread. **Also in this PR** - `research/optixtrans-e9612-stated-build.md` — the reading, the Table 3 rows, the gated ids. - `research/open-device-requests.json` — #808 removed; #810 narrowed: validity is now stated (Table 3 Y in both modes), so only the cross-connect mode is missing (1:1 -> 240 ports, 1:3 -> 200). - Generated pages regenerated (`regen:registry-doc`): 2,355 models / 258 vendors / 29,481 sockets / 971 shapes. **Gates:** `syntax-check` 411 files OK · `audit-port-types` no contradiction · `hardware-invariants` 26/26 · `import-devices-capabilities` 34/34 · `catalogue-hygiene` 25/25 · `faceplate-catalogue` 14/14 · `fuzz-importer` 14/14 · `importer-roundtrip` 26/26 · `doc-catalog-counts` 43/43 · `check-control-bytes` clean · `regen-registry-doc --check` current.
huawei: model the OptiXtrans E9612 as a stated 13x TNG1N210 build (#808)
Some checks failed
CI / Classify the change (pull_request) Successful in 6s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 9s
CI / Server smoke test (pull_request) Successful in 7s
CI / Docker build (pull_request) Successful in 20s
CI / Test (Node 20) (pull_request) Failing after 45s
CI / Test (Node 18) (pull_request) Failing after 54s
CI / Test (Node 22) (pull_request) Failing after 50s
CI / Arm auto-merge (pull_request) Successful in 4s
8a314bbc60
The E9600 Boards Overview Table 3 prints TNG1N206/TNG1N210 as Y in the E9612
column with no mode qualifier, and Table 1 names the board 10 x 10 Gbit/s; the
vendor spec table gives the E9612 13 service slots and Cross-connect Type NA.
13 x 10 = 130 line ports, no inference beyond that arithmetic.

The 13 is conditional (DC 13-slot layout; DC master 12; AC/HVDC 10-11 because
one APIU takes IU1 and IU2) and the spec notes say so. The board chapter is
gated (EDOC1100126589BEJ09047 / EDOC1100163901 -> 302 /view/notFound), so the
notes record that the Table 1/3 rows were read from the verbatim partner copy
and name the two vendor fields that came from the product page.

Research: research/optixtrans-e9612-stated-build.md. Registry regenerated from
research/open-device-requests.json (808 removed, 810 narrowed to the mode).
jaylouisw force-pushed catalogue/1118-0856 from 8a314bbc60
Some checks failed
CI / Classify the change (pull_request) Successful in 6s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 9s
CI / Server smoke test (pull_request) Successful in 7s
CI / Docker build (pull_request) Successful in 20s
CI / Test (Node 20) (pull_request) Failing after 45s
CI / Test (Node 18) (pull_request) Failing after 54s
CI / Test (Node 22) (pull_request) Failing after 50s
CI / Arm auto-merge (pull_request) Successful in 4s
to ec3b431fec
Some checks failed
CI / Classify the change (pull_request) Successful in 7s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 5s
CI / Server smoke test (pull_request) Successful in 6s
CI / Docker build (pull_request) Successful in 20s
CI / Test (Node 20) (pull_request) Failing after 48s
CI / Test (Node 18) (pull_request) Failing after 53s
CI / Test (Node 22) (pull_request) Failing after 50s
CI / Arm auto-merge (pull_request) Successful in 2s
2026-11-18 09:02:16 +00:00
Compare
Author
Owner

Surgery loop 08:55 — the red is one guard line, not this spec. Reproduced on ec3b431f.

Head ec3b431f, 0 behind origin/main (9d852ed8), mergeable=true, 8 files (+894/-30). Docker build, smoke, classify, rebase and arm-auto-merge all success. The three Node legs fail on the same assertion:

node test/close-merged-issues.test.js
  not ok - none of the six open requests is in the catalogue,
           so none is closed by this workflow  [#808]
72 passed, 1 failed

test/close-merged-issues.test.js:305-318 still carries the literal queue that origin/main has, including:

[808, '[Device]: OptiXtrans E9612'],

and asserts none of the six has a spec. This PR correctly gives #808 one, so the premise the guard encodes is now false — the spec is right, the expectation is stale.

The fix is the two-line edit #1955 already made for #807 in 1304b0bd: drop the [808, ...] row and add the positive assertion that #808 resolves to devices/huawei/optixtrans-e9612.json (that commit is the exact template — same file, same shape). With that, this PR is mergeable and green; nothing in the spec needs to change.

Not mine to make: test/ is off-limits to this loop, so this needs the author or Jay.

Two things worth knowing before that lands. #1934 also models an E9612 (same 13x TNG1N210 / 130x 10G, different metadata: model Title-Case, plus color/icon, different prose and sourceUrl) — whichever of the two merges first leaves the other a duplicate of the same box. And this guard is a hand-maintained mirror of research/open-device-requests.json, so every device PR that resolves a queued request trips it; #1955 edited it, this PR and #1934 did not. Making that assertion read the queue and the catalogue instead of a literal list would retire the trap.

**Surgery loop 08:55 — the red is one guard line, not this spec. Reproduced on `ec3b431f`.** Head `ec3b431f`, **0 behind `origin/main`** (`9d852ed8`), `mergeable=true`, 8 files (+894/-30). Docker build, smoke, classify, rebase and arm-auto-merge all success. The three Node legs fail on the same assertion: ``` node test/close-merged-issues.test.js not ok - none of the six open requests is in the catalogue, so none is closed by this workflow [#808] 72 passed, 1 failed ``` `test/close-merged-issues.test.js:305-318` still carries the literal queue that `origin/main` has, including: ```js [808, '[Device]: OptiXtrans E9612'], ``` and asserts none of the six has a spec. This PR correctly gives #808 one, so the premise the guard encodes is now false — the spec is right, the expectation is stale. **The fix is the two-line edit #1955 already made for #807 in `1304b0bd`:** drop the `[808, ...]` row and add the positive assertion that #808 resolves to `devices/huawei/optixtrans-e9612.json` (that commit is the exact template — same file, same shape). With that, this PR is mergeable and green; nothing in the spec needs to change. Not mine to make: `test/` is off-limits to this loop, so this needs the author or Jay. **Two things worth knowing before that lands.** `#1934` also models an E9612 (same 13x TNG1N210 / 130x 10G, different metadata: `model` Title-Case, plus `color`/`icon`, different prose and `sourceUrl`) — whichever of the two merges first leaves the other a duplicate of the same box. And this guard is a hand-maintained mirror of `research/open-device-requests.json`, so *every* device PR that resolves a queued request trips it; #1955 edited it, this PR and #1934 did not. Making that assertion read the queue and the catalogue instead of a literal list would retire the trap.
Author
Owner

Re-verified at ec3b431f merged with origin/main 9d852ed8 (2026-11-18 09:1x): node test/close-merged-issues.test.js → 72 passed, 1 failed, the single failure being

not ok - none of the six open requests is in the catalogue, so none is closed by this workflow  [#808]

Nothing else in the suite fails, and this PR is still 0 commits behind main with mergeable: true — so the guard literal is the only thing between it and a green run. The blocker is now tracked as #1963 rather than living only in this thread, so it survives this PR. No action needed here beyond that.

Re-verified at `ec3b431f` merged with `origin/main` `9d852ed8` (2026-11-18 09:1x): `node test/close-merged-issues.test.js` → **72 passed, 1 failed**, the single failure being ``` not ok - none of the six open requests is in the catalogue, so none is closed by this workflow [#808] ``` Nothing else in the suite fails, and this PR is still 0 commits behind main with `mergeable: true` — so the guard literal is the only thing between it and a green run. The blocker is now tracked as #1963 rather than living only in this thread, so it survives this PR. No action needed here beyond that.
test: E9612 is in the catalogue, so #808 leaves the unmodelled queue
All checks were successful
CI / Classify the change (pull_request) Successful in 7s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 8s
CI / Server smoke test (pull_request) Successful in 6s
CI / Docker build (pull_request) Successful in 20s
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 21s
19e13932e6
The real-catalogue block pins the open [Device]: queue and asserts none of
them is in the catalogue, because the close-merged-issues sweep would
otherwise close a request the catalogue does not answer. Adding
devices/huawei/optixtrans-e9612.json made that assertion false, which is
the honest signal that the request is now answered - so #808 moves out of
the queue the way #807's E9605 did when it landed, and gains its own
positive assertion that it resolves to its own file rather than E9624 or
E9605. No engine or sweep behaviour changes.
jaylouisw force-pushed catalogue/1118-0856 from 19e13932e6
All checks were successful
CI / Classify the change (pull_request) Successful in 7s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 8s
CI / Server smoke test (pull_request) Successful in 6s
CI / Docker build (pull_request) Successful in 20s
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 21s
to dfb22e0ae9
All checks were successful
CI / Classify the change (pull_request) Successful in 7s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 7s
CI / Server smoke test (pull_request) Successful in 13s
CI / Test (Node 20) (pull_request) Successful in 1m1s
CI / Test (Node 18) (pull_request) Successful in 1m5s
CI / Docker build (pull_request) Successful in 41s
CI / Test (Node 22) (pull_request) Successful in 1m4s
CI / Arm auto-merge (pull_request) Successful in 24s
2026-11-18 09:36:09 +00:00
Compare
Author
Owner

Surgery loop 09:23 — the blocker is now a five-line patch, verified green. It is Jay's call to apply, not mine.

I re-ran this at the current head against origin/main = 3aa97d60 (#1964 merged 09:1x). ec3b431f merged with main: same single failure, nothing else.

node test/close-merged-issues.test.js   ->  72 passed, 1 failed
  not ok - none of the six open requests is in the catalogue, so none is closed by this workflow  [#808]

Why the guard fails is not a defect in this spec, and not a decision about behaviour. test/close-merged-issues.test.js:306-317 hardcodes the open-request queue and asserts the catalogue answers none of it. Landing E9612 makes #808 answerable, so the assertion is supposed to fail — exactly as #807/E9605 did, and exactly how #1961 handled it (moved E9605 out of the queue into an "answered" assertion). This PR just never carried its half of that bookkeeping.

The patch, applied locally and reverted — not pushed:

diff --git a/test/close-merged-issues.test.js b/test/close-merged-issues.test.js
index ebc5db0f..4e8c6275 100644
--- a/test/close-merged-issues.test.js
+++ b/test/close-merged-issues.test.js
@@ -305,7 +305,6 @@ const INDEX = buildDeviceIndex(SPECS);
     const index = loadDeviceIndex();
     const queue = [
       [810, '[Device]: OptiXtrans E9624'],
-      [808, '[Device]: OptiXtrans E9612'],
       [803, '[Device]: OptiXtrans E6616X'],
       [743, '[Device]: Huawei OceanProtect X8100/X9100 Backup Storage'],
       [742, '[Device]: Huawei OceanProtect E8000 Backup Storage'],
@@ -313,13 +312,17 @@ const INDEX = buildDeviceIndex(SPECS);
     ];
     const proposed = queue.filter(function (q) { return deviceMatch(q[1], index); });
     ok(proposed.length === 0,
-      'none of the six open requests is in the catalogue, so none is closed by this workflow',
+      'none of the five open requests is in the catalogue, so none is closed by this workflow',
       proposed.map(function (p) { return '#' + p[0]; }).join(', '));
     // A request the catalogue now answers is answered: #807's E9605 landed on 2026-11-18, so the
     // sweep would find it - which is why it is no longer in the queue above.
     const answered = deviceMatch('[Device]: OptiXtrans E9605', index);
     ok(answered && answered.file === 'devices/huawei/optixtrans-e9605.json',
       'and the E9605 request that has since been built resolves to its spec file', answered && answered.file);
+    // #808's E9612 landed in the same wave, so it leaves the queue the same way.
+    const answered9612 = deviceMatch('[Device]: OptiXtrans E9612', index);
+    ok(answered9612 && answered9612.file === 'devices/huawei/optixtrans-e9612.json',
+      'and the E9612 request this PR answers resolves to its spec file', answered9612 && answered9612.file);
     // A near miss is a miss, and this is the near miss: E6616 is modelled, E6616X is not.
     ok(deviceMatch('[Device]: OptiXtrans E6616', index) !== null, 'while E6616 itself is modelled');
     ok(deviceMatch('[Device]: OptiXtrans E6616', index).file === 'devices/huawei/optixtrans-e6616.json',

With it applied in a scratch worktree at ec3b431f + 3aa97d60:

  • node test/close-merged-issues.test.js → 74 passed, 0 failed
  • full npm test → 338 suites in 473.9 s, 11,759 assertions, 0 failed (EXIT=0)

That is the entire distance between this PR and a green head — no source change, no semantic conflict, no other failing suite.

Jay: one word and I push this onto catalogue/1118-0856 (as a separate commit, authored by me, test/-only, --force-with-lease), which should turn all three Node legs green and let the standing automerge rule take it. I have not touched test/ on my own initiative — that block is still yours from the 08:55 escalation. The same patch is what #1934 will need, but only after the E9605 source-of-truth call.

The durable fix is still open as #1963 (derive the queue from research/open-device-requests.json); this patch is the stopgap for the two PRs in front of us.

**Surgery loop 09:23 — the blocker is now a five-line patch, verified green. It is Jay's call to apply, not mine.** I re-ran this at the current head against `origin/main` = `3aa97d60` (#1964 merged 09:1x). `ec3b431f` merged with main: same single failure, nothing else. ``` node test/close-merged-issues.test.js -> 72 passed, 1 failed not ok - none of the six open requests is in the catalogue, so none is closed by this workflow [#808] ``` **Why the guard fails is not a defect in this spec, and not a decision about behaviour.** `test/close-merged-issues.test.js:306-317` hardcodes the open-request queue and asserts the catalogue answers none of it. Landing E9612 makes `#808` answerable, so the assertion is *supposed* to fail — exactly as `#807`/E9605 did, and exactly how `#1961` handled it (moved E9605 out of the queue into an "answered" assertion). This PR just never carried its half of that bookkeeping. **The patch, applied locally and reverted — not pushed:** ```diff diff --git a/test/close-merged-issues.test.js b/test/close-merged-issues.test.js index ebc5db0f..4e8c6275 100644 --- a/test/close-merged-issues.test.js +++ b/test/close-merged-issues.test.js @@ -305,7 +305,6 @@ const INDEX = buildDeviceIndex(SPECS); const index = loadDeviceIndex(); const queue = [ [810, '[Device]: OptiXtrans E9624'], - [808, '[Device]: OptiXtrans E9612'], [803, '[Device]: OptiXtrans E6616X'], [743, '[Device]: Huawei OceanProtect X8100/X9100 Backup Storage'], [742, '[Device]: Huawei OceanProtect E8000 Backup Storage'], @@ -313,13 +312,17 @@ const INDEX = buildDeviceIndex(SPECS); ]; const proposed = queue.filter(function (q) { return deviceMatch(q[1], index); }); ok(proposed.length === 0, - 'none of the six open requests is in the catalogue, so none is closed by this workflow', + 'none of the five open requests is in the catalogue, so none is closed by this workflow', proposed.map(function (p) { return '#' + p[0]; }).join(', ')); // A request the catalogue now answers is answered: #807's E9605 landed on 2026-11-18, so the // sweep would find it - which is why it is no longer in the queue above. const answered = deviceMatch('[Device]: OptiXtrans E9605', index); ok(answered && answered.file === 'devices/huawei/optixtrans-e9605.json', 'and the E9605 request that has since been built resolves to its spec file', answered && answered.file); + // #808's E9612 landed in the same wave, so it leaves the queue the same way. + const answered9612 = deviceMatch('[Device]: OptiXtrans E9612', index); + ok(answered9612 && answered9612.file === 'devices/huawei/optixtrans-e9612.json', + 'and the E9612 request this PR answers resolves to its spec file', answered9612 && answered9612.file); // A near miss is a miss, and this is the near miss: E6616 is modelled, E6616X is not. ok(deviceMatch('[Device]: OptiXtrans E6616', index) !== null, 'while E6616 itself is modelled'); ok(deviceMatch('[Device]: OptiXtrans E6616', index).file === 'devices/huawei/optixtrans-e6616.json', ``` With it applied in a scratch worktree at `ec3b431f` + `3aa97d60`: - `node test/close-merged-issues.test.js` → **74 passed, 0 failed** - full `npm test` → **338 suites in 473.9 s, 11,759 assertions, 0 failed** (`EXIT=0`) That is the entire distance between this PR and a green head — no source change, no semantic conflict, no other failing suite. **Jay:** one word and I push this onto `catalogue/1118-0856` (as a separate commit, authored by me, `test/`-only, `--force-with-lease`), which should turn all three Node legs green and let the standing automerge rule take it. I have not touched `test/` on my own initiative — that block is still yours from the 08:55 escalation. The same patch is what `#1934` will need, but only after the E9605 source-of-truth call. The durable fix is still open as `#1963` (derive the queue from `research/open-device-requests.json`); this patch is the stopgap for the two PRs in front of us.
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!1962
No description provided.