ci(#6655): increase dispatchWait from 12 to 18 minutes - #6657
ci(#6655): increase dispatchWait from 12 to 18 minutes#6657fullsend-ai-coder[bot] wants to merge 1 commit into
Conversation
Fork PR behaviour tests fail intermittently because workflows complete successfully but exceed the 12-minute dispatchWait timeout. For example, PR #6611 had a workflow that completed in 12m16s — 16 seconds after the test gave up waiting. Increase dispatchWait to 18 minutes (50% headroom) to accommodate normal variance in GitHub Actions execution time while still catching genuine hangs. The dispatchMaxTry count (48) is unchanged — only the wait ceiling changes. Closes #6655
|
🤖 Finished Review · ✅ Success · Started 7:49 PM UTC · Completed 8:00 PM UTC Commit: Runtime: claude · Model: opus → claude-opus-4-6 · Effort: high · Cost: $2.52 |
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
|
Risk Assessment: moderate (2/5) DetailsSingle-line timeout constant bump in test infrastructure with perfect issue alignment and bot author, but the file's high recent churn (11 commits/30d) and regression history (8 fix/revert commits/90d) elevate risk slightly above low. |
|
Looks good to me |
![fullsend-ai-review[bot]](https://avatars.githubusercontent.com/in/3478501?s=60&v=4){kind=small}
fullsend-ai-review
Bot
added
the
ready-for-merge
|
In #6656, I figured out why they're so slow. Let's fix that, rather than extend the timeout. |
|
🤖 Finished Retro · ✅ Success · Started 8:13 PM UTC · Completed 8:23 PM UTC Commit: Runtime: claude · Model: opus → claude-opus-4-6 · Effort: high · Cost: $3.79 |
Retro: PR #6657 — Timeout extension closed in favor of root-cause fixWhat happenedralphbean created two companion issues simultaneously:
The code agent (run 33006220760, $1.73) faithfully implemented the prescribed single-line change in Meanwhile, ralphbean investigated #6656 and discovered the root cause: test pool repos accumulate ~1.9GB of git history from repeated runs without cleanup, causing the pre-review shallow-clone deepening step to take 12+ minutes. He closed the PR without merging, commenting: "Let's fix that, rather than extend the timeout." AnalysisCode agent performed well within its scope — the issue was highly prescriptive, the change was clean, conventions were followed, and cost was low ($1.73, 3 minutes wall time). The agent cannot be faulted for implementing exactly what was asked. Review agent approved a symptomatic fix without surfacing the higher-order concern. The triage agent's comment on #6655 explicitly noted "#6656 tracks root cause investigation into why fork PR workflows are taking 12+ minutes." A human reviewer in this position would likely have said "this looks correct as a tactical fix, but consider whether it's still needed given the root-cause investigation in #6656." The review agent's intent-coherence sub-agent had access to this context but did not flag the relationship. Total cost of discarded work: $4.25 (code + review). This is modest and partly reflects the natural cost of parallel exploration — the human was deliberately pursuing both tactical and root-cause tracks simultaneously. Existing issue coverageThe improvement opportunities from this retro are well-covered by existing open issues. New evidence from this case:
No new proposals are warranted — the existing issues comprehensively cover the gaps observed in this workflow. |
1 participant
Summary
dispatchWaittimeout from 12 minutes to 18 minutes (50% headroom) inpkg/behaviourtest/drivers/ci/githubactions/githubactions.godispatchMaxTrycount (48) is unchanged — only the wait ceiling changesTesting
pkg/behaviourtest/drivers/ci/githubactions/passgo vetpassesCloses #6655
Post-script verification
agent/6655-increase-dispatch-wait)d0d567bb7f882f797898e1d478948a4d82bfd1d6..HEAD)