devices(tp-link): the SG2008P's four PoE+ ports sourced nothing, and its sibling invented four #2377

Merged
jaylouisw merged 1 commit from catalogue/1121-0552 into main 2026-11-21 06:09:58 +00:00
Owner

Refs #130 (not a close: the Dream Wall Pro request is still blocked - see the comment on the issue, unchanged this run).

This run's increment is the catalogue defect underneath the request queue: TP-Link Omada SG2008P, the current 8-port PoE+ switch, was modelled as a PoE switch with no power supply at all.

devices/tp-link/SG2008P.json said in prose "4 PoE+ ports and 62 W power budget" and declared none of it: no poeOut on any port, no poe-pse capability, no poe block. show power inline answered % No PoE on ...: no port on this device can supply power. audit-poe-budget.js could not see it either - it counts devices that already source power somewhere, and this one sourced on none.

Vendor source (both the tp-link.com and omadanetworks.com paths for the SKU resolve to the same document, HTTP 200, 187,036 B, read 2026-11-21):
https://www.omadanetworks.com/us/business-networking/omada-switch-access/sg2008p/

  • Interface - 8x 10/100/1000 Mbps RJ45 Ports (Ports 1-4 Support 802.3at/af PoE+)
  • PoE Ports (RJ45) - Standard: 802.3af/at compliant / PoE Ports: 4 Ports / Power Supply: 62 W*
  • Power Supply - AC/DC Adapter (Input: 100-240V AC; Output: 53.5V DC/1.31A) = 70.1 W
  • Max. Power Consumption - 73.4 W (220 V/50 Hz @ 25 C) (with 62 W PD connected)

The 62 W is the PD-side allowance, not the adapter's rating: the max-consumption row is 62 W of PD load plus ~11 W of chassis, which only works if 62 W is what the PoE ports may deliver. Four 802.3af loads (61.6 W) fit; four Class-4 25.5 W loads (102 W) do not.

Also fixed, same family: devices/tp-link/tl-sg2008p.json (end-of-sale sibling, identical 8-port / four-PoE+ table) had the other half of the defect - four real PSE ports, no budget, and four invented ones, because engine/media.js fills a silent poeOut on every RJ45 port of a switch claiming poe-pse. eth4-eth7 were credited 15.4 W apiece. Both rows now declare ports 1-4 PoE+ and ports 5-8 poeOut: false, and carry the vendor's 62 W in poe.budgetW.

No per-port ceiling is recorded: 30 W is the 802.3at standard, not a figure this vendor prints for this box.

Gates: syntax-check 516 files OK; audit-port-types gated 0 contradictions; audit-poe-budget silent 37 -> 36 (TL-SG2008P off the list, SG2008P now counted and funded); import-devices --dry-run clean; hardware-invariants 26/26; full suite 436 suites / 14,654 assertions, 0 failed. regen-registry-doc says the pages are current (2,352 models).

New test test/tp-link-sg2008p-poe.test.js (30 assertions): budget 62, exactly four PSE ports, show power inline prints Available:62.0(w), two Class-4 APs fit at 51 W and the third is refused with PSE budget exhausted: 25.5 W requested, 11 W left.

Evidence written up in research/tp-link-sg2008p-poe-budget.md.

Refs #130 (not a close: the Dream Wall Pro request is still blocked - see the comment on the issue, unchanged this run). This run's increment is the catalogue defect underneath the request queue: **TP-Link Omada SG2008P**, the current 8-port PoE+ switch, was modelled as a PoE switch with no power supply at all. `devices/tp-link/SG2008P.json` said in prose "4 PoE+ ports and 62 W power budget" and declared none of it: no `poeOut` on any port, no `poe-pse` capability, no `poe` block. `show power inline` answered `% No PoE on ...: no port on this device can supply power`. `audit-poe-budget.js` could not see it either - it counts devices that already source power somewhere, and this one sourced on none. **Vendor source** (both the tp-link.com and omadanetworks.com paths for the SKU resolve to the same document, HTTP 200, 187,036 B, read 2026-11-21): https://www.omadanetworks.com/us/business-networking/omada-switch-access/sg2008p/ * `Interface - 8x 10/100/1000 Mbps RJ45 Ports (Ports 1-4 Support 802.3at/af PoE+)` * `PoE Ports (RJ45) - Standard: 802.3af/at compliant / PoE Ports: 4 Ports / Power Supply: 62 W*` * `Power Supply - AC/DC Adapter (Input: 100-240V AC; Output: 53.5V DC/1.31A)` = 70.1 W * `Max. Power Consumption - 73.4 W (220 V/50 Hz @ 25 C) (with 62 W PD connected)` The 62 W is the PD-side allowance, not the adapter's rating: the max-consumption row is 62 W of PD load plus ~11 W of chassis, which only works if 62 W is what the PoE ports may deliver. Four 802.3af loads (61.6 W) fit; four Class-4 25.5 W loads (102 W) do not. **Also fixed, same family:** `devices/tp-link/tl-sg2008p.json` (end-of-sale sibling, identical 8-port / four-PoE+ table) had the other half of the defect - four real PSE ports, no budget, and four *invented* ones, because `engine/media.js` fills a silent `poeOut` on every RJ45 port of a `switch` claiming `poe-pse`. `eth4-eth7` were credited 15.4 W apiece. Both rows now declare ports 1-4 PoE+ and ports 5-8 `poeOut: false`, and carry the vendor's 62 W in `poe.budgetW`. No per-port ceiling is recorded: 30 W is the 802.3at standard, not a figure this vendor prints for this box. **Gates:** `syntax-check` 516 files OK; `audit-port-types` gated 0 contradictions; `audit-poe-budget` silent 37 -> 36 (TL-SG2008P off the list, SG2008P now counted and funded); `import-devices --dry-run` clean; `hardware-invariants` 26/26; full suite 436 suites / 14,654 assertions, 0 failed. `regen-registry-doc` says the pages are current (2,352 models). New test `test/tp-link-sg2008p-poe.test.js` (30 assertions): budget 62, exactly four PSE ports, `show power inline` prints `Available:62.0(w)`, two Class-4 APs fit at 51 W and the third is refused with `PSE budget exhausted: 25.5 W requested, 11 W left`. Evidence written up in `research/tp-link-sg2008p-poe-budget.md`.
devices(tp-link): the SG2008P's PoE ports sourced nothing, and its sibling invented four (refs #130)
All checks were successful
CI / Classify the change (pull_request) Successful in 14s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 19s
CI / Test (Node 22) (pull_request) Successful in 1m5s
CI / Test (Node 18) (pull_request) Successful in 1m5s
CI / Test (Node 20) (pull_request) Successful in 1m7s
CI / Server smoke and Docker build (pull_request) Successful in 35s
CI / Arm auto-merge (pull_request) Successful in 5s
43e89c02ee
forgejo-actions scheduled this pull request to auto merge when all checks succeed 2026-11-21 06:02:25 +00:00
jaylouisw deleted branch catalogue/1121-0552 2026-11-21 06:10:00 +00:00
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
jaylouisw/packetforge!2377
No description provided.