fix(web): register CLI stop signals before printing readiness - #401
Open
tt-a1i wants to merge 3 commits into
Open
fix(web): register CLI stop signals before printing readiness#401tt-a1i wants to merge 3 commits into
tt-a1i wants to merge 3 commits into
Conversation
Register SIGINT/SIGTERM with process.once after host.start and before readiness so a SIGTERM at ready cannot miss the handler. Keep once so a second same signal can still default-terminate while stop is in flight.
waitForPath only stats existence, so a direct writeFile can let the parent observe an empty stop-entered file before contents land.
Directory watch can fire first for the staging rename and miss the final name. Record first SIGTERM delivery and inspect both paths on timeout.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Register SIGINT/SIGTERM with
process.onceimmediately afterhost.start(), before readiness output. Tests that send SIGTERM on the ready line can otherwise observeexitCode === null/signal === "SIGTERM"instead of the CLI stop path.Keep
once(noton). A second same signal whilestop()is in flight falls through to Node default termination.Tests:
process.once(signal, onStopSignal)is registered before readiness output.exitCode === 1,signal === null, stderr includesstop failed. Unchanged.enteredvia same-dir write-then-rename, then writesstop-enteredon stderr. Parent waits for file contents, woken by directory watch or that stderr, records firstkill()delivery, and on timeout inspects final + staging paths.waitForPathexistence-only is gone.Does not include the #355 peer-entry resolver. Does not land inside #373. Production
bin/openpi.jsis unchanged by the hang-stop follow-ups.Test plan
bun run check;tests/web/cli.test.ts4/4 on Node 26.3.0 and portable Node 22.19.0 Darwinstop-entered.tmpthenstop-enteredbun run test: 1340 / pass 1337 / fail 2 / skip 1. Hang-stop passed. Failures are plan-mode result-rendering only. Vitest skipped by runner; separate official Vitest 30/30 EXIT=06e7ca46https://github.com/openpi-dev/openpi/actions/runs/33943551441 : Node 22.19.0 / Node 24 / Windows all pass. Node 22 hang-stopok 1222; Node 22 packed standalone Web CLI smoke ran and the job completed (nokill: No such process). Windows hang-stop is a no-op return (ok 1129)Pascal's 1/10
stopMarker5s timeout with a live child is retained as fact. It is not attributed to signal-before-handler: production registersprocess.oncebefore readiness. Whether that timeout was a missed rename watch or first SIGTERM never enteringstop()is unknown.