poe: a refused port is power-deny, and draws 0.0 W #2149

Merged
jaylouisw merged 1 commit from rev/1119-1212 into main 2026-11-19 12:39:55 +00:00
Owner

One question: when the PSE detects a powered device and cannot supply it, what does the Oper column say, and what sits in the power cell?

Verdict: power-deny, 0.0. Oper has four documented values (on, off, faulty, power-deny); off means "No PoE is applied" - the word an empty socket prints. This console printed off for a refusal, plus a (denied) tag and the device's request in the allocated-power cell.

Sources (all in research/show-power-inline-power-deny.md):

  • Catalyst 2960-X 15.2(7)E power inline - "the PoE port is in a power-deny state ... the Oper column in the show power inline privileged EXEC command output shows power-deny".
  • 2960-X 15.2(2)E / 15.2(5)E, 3850 16.1, 9300 17.12 - the same Oper field table and the sample row Gi3/0/1 auto power-deny 4.0 n/a n/a 15.4.
  • Budget-exhausted captures, which is the only denial this model has: IE-4000 thread Gi1/13 auto power-deny 0.0 n/a n/a 30.0, Catalyst 3650 Gi1/0/9 auto power-deny 0.0 n/a n/a 30.0, Cisco Learning Network Gi3/0/22 auto power-deny 0.0 n/a n/a 30.0.

Changes:

  • engine/cli.js - refused rows carry Oper = power-deny and Draw = 0.0 W; the (denied) tag is dropped because the word now says it, and the denial sentence under the table (... PSE budget exhausted: 12.95 W requested, 12.05 W left) still carries the request. Oper widened to 11 characters, the width the box's own heading holds, so the rest of the row stays in line.
  • test/show-power-inline.test.js - the refused row is pinned cell by cell, with the neighbouring powered / non-PD / shut rows pinned to on / off / off so the new word cannot leak.
  • test/poe-allocation-order.test.js - the lowest port that does not fit now reads power-deny.
  • docs/CLI.md - sample table and the bullet on the denied row.

npm test: 373 suites, 12,682 assertions, 0 failed. node scripts/syntax-check.js: 452 files OK.

Still open (recorded in the note): faulty is unreachable (no fault path), the power inline police view is not modelled, and the Class/Max columns are a separate layout question.

One question: when the PSE detects a powered device and cannot supply it, what does the `Oper` column say, and what sits in the power cell? **Verdict: `power-deny`, `0.0`.** `Oper` has four documented values (`on`, `off`, `faulty`, `power-deny`); `off` means "No PoE is applied" - the word an empty socket prints. This console printed `off` for a refusal, plus a `(denied)` tag and the device's request in the allocated-power cell. Sources (all in `research/show-power-inline-power-deny.md`): * Catalyst 2960-X 15.2(7)E `power inline` - "the PoE port is in a power-deny state ... the Oper column in the show power inline privileged EXEC command output shows power-deny". * 2960-X 15.2(2)E / 15.2(5)E, 3850 16.1, 9300 17.12 - the same `Oper` field table and the sample row `Gi3/0/1 auto power-deny 4.0 n/a n/a 15.4`. * Budget-exhausted captures, which is the only denial this model has: IE-4000 thread `Gi1/13 auto power-deny 0.0 n/a n/a 30.0`, Catalyst 3650 `Gi1/0/9 auto power-deny 0.0 n/a n/a 30.0`, Cisco Learning Network `Gi3/0/22 auto power-deny 0.0 n/a n/a 30.0`. Changes: * `engine/cli.js` - refused rows carry `Oper = power-deny` and `Draw = 0.0 W`; the `(denied)` tag is dropped because the word now says it, and the denial sentence under the table (`... PSE budget exhausted: 12.95 W requested, 12.05 W left`) still carries the request. `Oper` widened to 11 characters, the width the box's own heading holds, so the rest of the row stays in line. * `test/show-power-inline.test.js` - the refused row is pinned cell by cell, with the neighbouring powered / non-PD / shut rows pinned to `on` / `off` / `off` so the new word cannot leak. * `test/poe-allocation-order.test.js` - the lowest port that does not fit now reads `power-deny`. * `docs/CLI.md` - sample table and the bullet on the denied row. `npm test`: 373 suites, 12,682 assertions, 0 failed. `node scripts/syntax-check.js`: 452 files OK. Still open (recorded in the note): `faulty` is unreachable (no fault path), the `power inline police` view is not modelled, and the `Class`/`Max` columns are a separate layout question.
poe: a refused port is power-deny, and draws 0.0 W
All checks were successful
CI / Classify the change (pull_request) Successful in 16s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 6s
CI / Server smoke test (pull_request) Successful in 10s
CI / Test (Node 22) (pull_request) Successful in 56s
CI / Test (Node 20) (pull_request) Successful in 56s
CI / Docker build (pull_request) Successful in 20s
CI / Test (Node 18) (pull_request) Successful in 1m0s
CI / Arm auto-merge (pull_request) Successful in 3s
678c35a459
`Oper` in `show power inline` has four documented values - on, off, faulty,
power-deny - and `off` is defined as "No PoE is applied", the word a socket
with nothing on it prints. A port that detected a powered device and could not
supply it was printing `off`, so the one row an operator is looking for read
exactly like an empty port, with a local `(denied)` tag and the device's
12.95 W request priced in a cell that is meant to hold the power allocated.

Cisco's `power inline` reference states the rule in one sentence ("the Oper
column ... shows power-deny"), four command references carry the same field
table, and budget-exhausted captures from an IE-4000, a Catalyst 3650 and a
learning-network thread show the row as `auto power-deny 0.0 n/a n/a 30.0`.
The row now reads `auto power-deny 0.0 W PDB`; the requested wattage stays in
the denial sentence under the table, which is unchanged. The `Oper` field is
eleven characters, the width the box's heading holds, so the columns behind it
stay in line.

research/show-power-inline-power-deny.md carries the sources and what is still
open (`faulty`, the policing view, the Class/Max columns).
forgejo-actions scheduled this pull request to auto merge when all checks succeed 2026-11-19 12:25:21 +00:00
jaylouisw deleted branch rev/1119-1212 2026-11-19 12:39:58 +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!2149
No description provided.