fix(reply): let preflight compaction use compaction timeout by yu-xin-c · Pull Request #95590 · openclaw/openclaw

yu-xin-c · 2026-06-21T16:18:12Z

What Problem This Solves

Fixes #95553. Required preflight compaction could inherit the outer reply-operation abort signal, so a short reply lifecycle timeout could cancel compaction before the configured compaction timeout had a chance to govern the required preflight work.

Why This Change Was Made

This keeps required preflight compaction bounded by the compaction layer while still preserving real cancellation. ReplyOperation exposes an explicit cancellation signal for user stop and restart cancellation, and required preflight compaction now receives a scoped preflight abort signal that:

ignores upstream lifecycle TimeoutError aborts so the compaction timeout owns slow required compaction;
preserves upstream non-timeout aborts such as chat/RPC/user/restart cancellation;
preserves explicit stop/restart cancellation even if the lifecycle timeout has already aborted the main reply signal.

The patch is intentionally narrow: normal reply runs still use the existing operation abort signal, while required preflight compaction uses the filtered signal because it already has its own safety timeout.

User Impact

When a reply needs preflight compaction, users should see compaction complete or fail according to compaction.timeoutSeconds, instead of being prematurely canceled by the reply lifecycle timeout path. If the user explicitly stops the run, an RPC abort arrives, or the run is restarted, the preflight compaction still cancels promptly instead of waiting for the compaction timeout.

This PR is split out from #95563 to keep the P1 behavior fix reviewable on its own.

Evidence

node scripts/run-vitest.mjs src/auto-reply/reply/agent-runner-memory.test.ts src/auto-reply/reply/reply-run-registry.test.ts src/agents/embedded-agent-runner.compaction-safety-timeout.test.ts passed after rebasing onto origin/main: 2 auto-reply files / 68 tests and 1 agents file / 22 tests.
corepack pnpm tsgo:core passed after rebasing onto origin/main.
git diff --check passed.
2026-06-24 redacted production preflight proof: node --import tsx invoked the production runPreflightCompactionIfNeeded path with a real ReplyOperation, a slow compaction seam, a simulated lifecycle TimeoutError, and a real abortByUser() explicit cancellation. Redacted terminal output:

OpenClaw preflight compaction cancellation proof
paths redacted; no credentials or channel identifiers used
  notice=start
SCENARIO lifecycle-timeout-ignored
  preflight-start phase=preflight_compacting
  compact-signal-initially-aborted=false
  lifecycle-timeout-aborted-at-ms=261
  compact-signal-after-lifecycle-timeout=false
  slow-compaction-finished-at-ms=440
  compact-signal-at-finish=false
  compaction-count-incremented=true
  notice=end
  preflight-result=completed
  operation-phase-after=preflight_compacting
  notice=start
SCENARIO explicit-user-abort-cancels
  preflight-start phase=preflight_compacting
  compact-signal-initially-aborted=false
  explicit-user-abort-called-at-ms=83
  compact-signal-after-explicit-abort=true
  compact-signal-reason-name=AbortError
  explicit-abort-observed-at-ms=83
  notice=incomplete
  preflight-result=threw-AbortError
  operation-phase-after=aborted

Known remaining proof gap: I still have not added logs from the reporter's exact slow vLLM/Qwen backend. The new proof above exercises the production preflight entry and reply cancellation contract with a slow compaction seam, plus focused regression tests and typecheck.

clawsweeper · 2026-06-21T16:19:45Z

Codex review: needs maintainer review before merge. Reviewed June 24, 2026, 8:31 AM ET / 12:31 UTC.

Summary
The PR adds a scoped preflight compaction abort signal and an explicit reply cancellation signal so required budget/preflight compaction ignores lifecycle TimeoutError while still honoring explicit and non-timeout cancellation, with regression tests.

PR surface: Source +64, Tests +194. Total +258 across 4 files.

Reproducibility: yes. at source level: origin/main and v2026.6.10 still pass replyOperation.abortSignal into required preflight compaction, and compactWithSafetyTimeout races external aborts before the configured timeout. I did not run the reporter's exact slow vLLM/Qwen backend.

Review metrics: 1 noteworthy metric.

Reply Cancellation Surface: 1 optional signal added, 1 preflight signal source changed. The PR changes which abort events can interrupt required preflight compaction, so maintainers should verify timeout ownership and explicit cancellation before merge.

Root-cause cluster
Relationship: fixed_by_candidate
Canonical: #95553
Summary: The canonical remaining work is the preflight/budget compaction reply-abort timeout bug; this PR is an active candidate fix for the cancellation contract, while config-gate and partial-progress work overlap but do not replace it.

Members:

canonical: Bug: preflight (budget-triggered) compaction hard-capped at ~60s, ignores compaction.timeoutSeconds #95553 - The PR uses closing syntax for this issue, and the issue body describes the exact preflight compaction timeout problem this branch edits.
fixed_by_candidate: fix: preflight compaction uses reply abort signal (~60s) instead of configurable compaction timeout (#95553) #95561 - This open PR also targets the same preflight compaction abort-signal bug with a different signal-composition approach.
partial_overlap: feat(compaction): add compaction.preflight.enabled config gate (#95553) #95675 - This open PR adds an optional config gate for preflight compaction, addressing the issue's escape-hatch dimension rather than the cancellation-signal timeout path.
partial_overlap: Bug: 180s compaction timeout is a single wall clock over the whole chunk pipeline with no partial-progress reuse, so a legitimately-long compaction fails identically every turn #92043 - This issue tracks broader compaction timeout partial-progress and user-facing timeout guidance, overlapping symptoms but not the specific reply-operation abort source.
partial_overlap: fix(compaction): later provider timeouts discard completed chunks #92899 - This open PR targets broader compaction timeout partial-progress behavior and does not supersede the preflight reply-abort fix.

Proposal only: this assessment does not dispatch repair, suppress jobs, mutate sibling items, close, or merge anything.

Merge readiness
Overall: 🐚 platinum hermit
Proof: 🐚 platinum hermit
Patch quality: 🐚 platinum hermit
Result: ready for maintainer review.

Overall follows the weaker of proof and patch quality, so missing proof can cap an otherwise strong patch.

Rank-up moves:

[P2] Maintainers should choose whether this filtered-signal branch is the canonical cancellation fix before landing.

Risk before merge

[P2] Required preflight compaction runs while session context is near overflow, so a wrong cancellation contract could leave session state locked or stale until the configured compaction timeout.
[P1] The new terminal proof exercises the production preflight entry with a slow seam, but it still does not reproduce the reporter's exact slow vLLM/Qwen deployment.
[P1] Several open candidates target the same canonical bug or its config escape hatch, so maintainers still need to choose the final cancellation/config path before landing one branch.

Maintainer options:

Make This The Cancellation Fix (recommended)
Maintainers can accept this filtered-signal contract as the canonical fix if they agree lifecycle TimeoutError should be ignored only for required preflight compaction while explicit and non-timeout aborts still cancel.
Request Full Slow-Backend Proof
Maintainers can require an additional redacted run from the reporter's slow backend before merge if the production-path seam proof is not enough for this availability-sensitive path.
Resolve Competing Branches
After choosing the cancellation contract, close or retarget the overlapping abort-signal PRs and handle the config-gate PR as a separate product/config question.

Next step before merge

[P2] Maintainer review is needed to choose the canonical cancellation/config path and accept the session-state merge risk; no narrow automated repair is identified.

Security
Cleared: The diff changes TypeScript abort-signal plumbing and tests only; it adds no dependency, workflow, package, credential, network, or secret-handling surface.

Review details

Best possible solution:

Land one canonical cancellation-signal fix with this filtered-signal contract after maintainer review, and keep the optional preflight config gate as a separate decision.

Do we have a high-confidence way to reproduce the issue?

Yes at source level: origin/main and v2026.6.10 still pass replyOperation.abortSignal into required preflight compaction, and compactWithSafetyTimeout races external aborts before the configured timeout. I did not run the reporter's exact slow vLLM/Qwen backend.

Is this the best way to solve the issue?

Yes, the revised filtered-signal shape is the best fix I found for this PR's scope: it ignores only lifecycle TimeoutError while preserving explicit and upstream non-timeout cancellation. The optional preflight config gate should remain a separate maintainer decision.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against e4763b0631bf.

Label changes

Label justifications:

P1: The PR targets a large-session compaction regression that can block agent turns and hold session work in active deployments.
merge-risk: 🚨 session-state: Required preflight compaction operates on near-overflow transcript context, and incorrect cancellation can leave session context stale or locked longer than intended.
merge-risk: 🚨 availability: If cancellation is wrong, slow preflight compaction can keep a session unavailable until the configured compaction timeout.
rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🐚 platinum hermit and patch quality is 🐚 platinum hermit.
status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (terminal): The PR body now includes redacted terminal output from a production-path node run showing lifecycle timeout ignored and explicit user abort canceling the preflight signal after the fix.
proof: sufficient: Contributor real behavior proof is sufficient. The PR body now includes redacted terminal output from a production-path node run showing lifecycle timeout ignored and explicit user abort canceling the preflight signal after the fix.

Evidence reviewed

PR surface:

Source +64, Tests +194. Total +258 across 4 files.

View PR surface stats

Area	Files	Added	Removed	Net
Source	2	101	37	+64
Tests	2	196	2	+194
Docs	0	0	0	0
Config	0	0	0	0
Generated	0	0	0	0
Other	0	0	0	0
Total	4	297	39	+258

What I checked:

Root policy read: The full root AGENTS.md and the relevant src/agents/AGENTS.md were read; the review applied the repository proof bar and session-state merge-risk guidance. (AGENTS.md:1)
Current main still has the reported wiring: Current origin/main still passes params.replyOperation.abortSignal into required budget/preflight compaction, so the central bug is not already implemented on main. (src/auto-reply/reply/agent-runner-memory.ts:981, e4763b0631bf)
Compaction safety timeout races caller aborts: compactWithSafetyTimeout separately races any external abort signal against compaction, so the old preflight caller signal can win before the configured compaction timeout. (src/agents/embedded-agent-runner/compaction-safety-timeout.ts:104, e4763b0631bf)
Latest release has the old behavior: v2026.6.10 still contains the preflight abortSignal: params.replyOperation.abortSignal call, so the fix is not shipped in the latest release. (src/auto-reply/reply/agent-runner-memory.ts:981, aa69b12d0086)
PR head scopes the preflight abort signal: The PR creates a preflight abort signal that ignores timeout reasons from the reply operation signal while still watching explicit cancellation. (src/auto-reply/reply/agent-runner-memory.ts:93, 0607489dc8f4)
PR head adds explicit cancellation state: ReplyOperation gains explicitAbortSignal and abortWithReason aborts it for user/restart cancellation while upstream lifecycle abort remains on the main abortSignal. (src/auto-reply/reply/reply-run-registry.ts:58, 0607489dc8f4)

Likely related people:

jared596: Authored the merged PR that expanded runPreflightCompactionIfNeeded around transcript estimates, which is central to the affected preflight path. (role: preflight compaction feature contributor; confidence: high; commits: c6d8318d07f5; files: src/auto-reply/reply/agent-runner-memory.ts, src/auto-reply/reply/agent-runner.ts)
ArthurNie: Authored the merged fix that forced preflight compaction before oversized agent turns, directly adjacent to this required-preflight behavior. (role: preflight hard-gate contributor; confidence: high; commits: 9d54285b0d4a; files: src/auto-reply/reply/agent-runner-memory.ts)
dutifulbob: Authored the merged reply lifecycle unification across stop, rotation, and restart, which is the cancellation contract this PR changes. (role: reply lifecycle contributor; confidence: high; commits: 3f6840230b86; files: src/auto-reply/reply/reply-run-registry.ts)
wangmiao0668000666: Authored the merged compaction timeout default change that preserved explicit timeoutSeconds, which this PR tries to let preflight compaction honor. (role: timeout contract contributor; confidence: high; commits: bb6e47729cf8; files: src/agents/embedded-agent-runner/compaction-safety-timeout.ts, docs/gateway/config-agents.md, src/config/schema.help.ts)

What the crustacean ranks mean

🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

How this review workflow works

ClawSweeper keeps one durable marker-backed review comment per issue or PR.
Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
Maintainers can also comment @clawsweeper review to request a fresh review only.
Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

…law#95553) Adds a user-facing config gate to disable preflight (budget-triggered) compaction. When compaction.preflight.enabled === false, the pre-turn threshold check in runPreflightCompactionIfNeeded short-circuits to the existing session entry, so subsequent turns fall through to overflow recovery (which already honors compaction.timeoutSeconds with a 15min budget). The check is '=== false' against the literal value, so an unset key preserves the existing default-on behavior — the fix is strictly additive and does not change shipped behavior for any current user. This complements the four open PRs (openclaw#95561, openclaw#95563, openclaw#95580, openclaw#95590) that are all working on the abortSignal source for the preflight path. None of those PRs addresses the user-facing config dimension that the issue explicitly requests. Changes: - src/config/types.agent-defaults.ts: new AgentCompactionPreflightConfig type with optional enabled boolean - src/config/zod-schema.agent-defaults.ts: zod schema with strict mode - src/config/schema.help.ts + schema.labels.ts: help text + label - src/auto-reply/reply/agent-runner-memory.ts: 6-line early return in runPreflightCompactionIfNeeded, placed after isHeartbeat/isCli and before codex runtime so the gate is the only OpenClaw-specific path affected - src/auto-reply/reply/agent-runner-memory.test.ts: 1 new test (skips preflight compaction when enabled === false) using the same test fixture pattern as adjacent tests - scripts/repro/issue-95553-preflight-disabled-gate.mts: standalone real-environment proof that exercises the production gate Refs openclaw#95553

yu-xin-c · 2026-06-22T13:18:46Z

@clawsweeper re-review

clawsweeper · 2026-06-22T13:18:49Z

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event repository_dispatch).
Result: the existing ClawSweeper review comment will be edited in place when the review finishes.

yu-xin-c · 2026-06-23T15:04:02Z

@clawsweeper re-review

clawsweeper · 2026-06-23T15:04:06Z

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event repository_dispatch).
Result: the existing ClawSweeper review comment will be edited in place when the review finishes.

yu-xin-c · 2026-06-24T12:23:10Z

@clawsweeper re-review

openclaw-barnacle · 2026-07-11T04:26:40Z

This pull request has been automatically marked as stale due to inactivity.
Please add updates or it will be closed.

openclaw-barnacle · 2026-07-19T04:35:04Z

Closing due to inactivity.
If you believe this PR should be revived, post in #clawtributors on Discord to talk to a maintainer.
That channel is the escape hatch for high-quality PRs that get auto-closed.

openclaw-barnacle Bot added size: XS triage: needs-pr-context Candidate: external PR body lacks required problem context or evidence. labels Jun 21, 2026

yu-xin-c mentioned this pull request Jun 21, 2026

[codex] feat(mcp): share bundled runtime scope and fix preflight compaction abort #95563

Closed

openclaw-barnacle Bot removed the triage: needs-pr-context Candidate: external PR body lacks required problem context or evidence. label Jun 21, 2026

clawsweeper Bot mentioned this pull request Jun 21, 2026

fix(compaction): preflight compaction uses configured compaction timeout instead of reply operation abort signal #95580

Closed

23 tasks

This was referenced Jun 22, 2026

fix: preflight compaction uses reply abort signal (~60s) instead of configurable compaction timeout (#95553) #95561

Closed

feat(mcp): add shared bundled runtime scope #95591

Closed

yu-xin-c force-pushed the codex/95553-preflight-compaction-timeout branch from 2e201ff to 7aa7246 Compare June 22, 2026 13:07

openclaw-barnacle Bot added size: S and removed size: XS labels Jun 22, 2026

yu-xin-c added 3 commits June 23, 2026 22:56

fix(reply): let preflight compaction use compaction timeout

9beb370

fix(reply): preserve explicit preflight compaction cancellation

e3e6bed

fix(reply): preserve upstream preflight cancellation

0607489

yu-xin-c force-pushed the codex/95553-preflight-compaction-timeout branch from 7aa7246 to 0607489 Compare June 23, 2026 15:01

openclaw-barnacle Bot added size: M and removed size: S labels Jun 23, 2026

yu-xin-c changed the title ~~[codex] fix(reply): let preflight compaction use compaction timeout~~ fix(reply): let preflight compaction use compaction timeout Jun 23, 2026

yu-xin-c marked this pull request as ready for review June 24, 2026 13:07

bxy3045134656 mentioned this pull request Jun 26, 2026

Bug: Large session JSON file causes Gateway timeout — compaction cannot prevent it #96900

Closed

openclaw-barnacle Bot added the stale Marked as stale due to inactivity label Jul 11, 2026

openclaw-barnacle Bot closed this Jul 19, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

fix(reply): let preflight compaction use compaction timeout#95590

fix(reply): let preflight compaction use compaction timeout#95590
yu-xin-c wants to merge 3 commits into
openclaw:mainfrom
yu-xin-c:codex/95553-preflight-compaction-timeout

yu-xin-c commented Jun 21, 2026 •

edited

Loading

Uh oh!

clawsweeper Bot commented Jun 21, 2026 •

edited

Loading

Uh oh!

yu-xin-c commented Jun 22, 2026

Uh oh!

clawsweeper Bot commented Jun 22, 2026 •

edited

Loading

Uh oh!

yu-xin-c commented Jun 23, 2026

Uh oh!

clawsweeper Bot commented Jun 23, 2026 •

edited

Loading

Uh oh!

yu-xin-c commented Jun 24, 2026

Uh oh!

openclaw-barnacle Bot commented Jul 11, 2026

Uh oh!

openclaw-barnacle Bot commented Jul 19, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Uh oh!

Conversation

yu-xin-c commented Jun 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What Problem This Solves

Why This Change Was Made

User Impact

Evidence

Uh oh!

clawsweeper Bot commented Jun 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

yu-xin-c commented Jun 22, 2026

Uh oh!

clawsweeper Bot commented Jun 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

yu-xin-c commented Jun 23, 2026

Uh oh!

clawsweeper Bot commented Jun 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

yu-xin-c commented Jun 24, 2026

Uh oh!

openclaw-barnacle Bot commented Jul 11, 2026

Uh oh!

openclaw-barnacle Bot commented Jul 19, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

yu-xin-c commented Jun 21, 2026 •

edited

Loading

clawsweeper Bot commented Jun 21, 2026 •

edited

Loading

clawsweeper Bot commented Jun 22, 2026 •

edited

Loading

clawsweeper Bot commented Jun 23, 2026 •

edited

Loading