docs(convergence): the not-modelled list no longer denies the re-convergence it demonstrates #2028

Merged
jaylouisw merged 1 commit from docs/1118-2017 into main 2026-11-18 20:37:51 +00:00
Owner

docs/CONVERGENCE.md said, in "What this does not model", that "Changing a link after power-on and watching the network re-converge is not modelled; the schedule is computed from the topology as it stands", and that a cable pulled mid-convergence "is a separate feature". Both bullets date from #1050; #1058 landed exactly those two behaviours and added the "Re-convergence: a cable is pulled" section to this same page without pruning the list.

Evidence from the engine:

  • engine/convergence.js:149-200 — buildSchedule(topo, { fromMs }) re-bases every entry on fromMs and prepends a change entry at that instant.
  • engine/simulator.js:459 — getConvergence() passes fromMs: this.topo.lastChangeAtMs || 0.
  • Measured on two Catalyst 2960s cabled on l1 and l2: converge reports 30000 ms; link down l1 at 5000 answers "Network converges again at 35000 ms" and converge at 20000 shows change 5000 ms the link state changed with stp pending; link down l2 at 9000 answers 39000 ms and the change entry moves to 9000, the earlier deadline gone.

The two stale bullets become one true one: only the most recent change is kept, so a second pull re-bases the schedule rather than interleaving with the first.

Docs guards unchanged: syntax-check 422 files OK; docs-cli-transcript 4/0; docs-show-commands 7/0; no-form-help-key 81/0; help-completeness 8/0.

`docs/CONVERGENCE.md` said, in "What this does not model", that "Changing a link after power-on and watching the network re-converge is not modelled; the schedule is computed from the topology as it stands", and that a cable pulled mid-convergence "is a separate feature". Both bullets date from #1050; #1058 landed exactly those two behaviours and added the "Re-convergence: a cable is pulled" section to this same page without pruning the list. Evidence from the engine: - `engine/convergence.js:149-200` — `buildSchedule(topo, { fromMs })` re-bases every entry on `fromMs` and prepends a `change` entry at that instant. - `engine/simulator.js:459` — `getConvergence()` passes `fromMs: this.topo.lastChangeAtMs || 0`. - Measured on two Catalyst 2960s cabled on l1 and l2: `converge` reports 30000 ms; `link down l1 at 5000` answers "Network converges again at 35000 ms" and `converge at 20000` shows `change 5000 ms the link state changed` with stp pending; `link down l2 at 9000` answers 39000 ms and the change entry moves to 9000, the earlier deadline gone. The two stale bullets become one true one: only the most recent change is kept, so a second pull re-bases the schedule rather than interleaving with the first. Docs guards unchanged: syntax-check 422 files OK; docs-cli-transcript 4/0; docs-show-commands 7/0; no-form-help-key 81/0; help-completeness 8/0.
docs(convergence): the not-modelled list no longer denies the re-convergence it demonstrates
All checks were successful
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 4s
CI / Classify the change (pull_request) Successful in 5s
CI / Server smoke test (pull_request) Has been skipped
CI / Docker build (pull_request) Has been skipped
CI / Test (Node 20) (pull_request) Successful in 52s
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 26s
5bd4a839dc
The page's "What this does not model" list still carried two bullets from #1050:
"Topology-change driven convergence ... is not modelled" and "Link-down events
during convergence ... is a separate feature". #1058 landed both — the schedule
is re-based from topology.lastChangeAtMs (engine/convergence.js reads opts.fromMs,
Simulator.getConvergence passes topo.lastChangeAtMs) — and the page's own
"Re-convergence: a cable is pulled" section, plus LINK-STATE.md, describe the
behaviour. Two stale bullets become one true one: only the most recent change is
kept, so a second pull re-bases the schedule instead of interleaving with the
first.

Measured: two 2960s on l1 and l2 converge at 30000 ms; link down l1 at 5000 ->
"Network converges again at 35000 ms", and converge at 20000 shows the change
entry at 5000 with stp pending; link down l2 at 9000 -> 39000 ms, and the change
entry moves to 9000, the 35000 deadline gone.
forgejo-actions scheduled this pull request to auto merge when all checks succeed 2026-11-18 20:30:23 +00:00
jaylouisw deleted branch docs/1118-2017 2026-11-18 20:37:53 +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!2028
No description provided.