ci: bound the SwiftPM scratch holder and cache scratch sizes by teamleaderleo · Pull Request #15366 · manaflow-ai/cmux

teamleaderleo · 2026-09-28T12:57:07Z

A follow-up to #14804, from its review.

Bound the holder. owned_spm_scratch.py link starts a small process that holds a shared flock on the job's SwiftPM scratch directory. Until now it exited only when the runner killed a job's leftover processes at job end. If the runner process died mid-job, the holder kept the lock until the mini rebooted, and that directory could never be pruned or evicted. hold now exits after 65 minutes (signal.alarm), just past the swift-package-tests job's 60-minute timeout.

Stop walking the whole scratch on every link. tree_stats lstat'ed every file of up to 24 GiB of scratch to size and rank each directory. Now:

link touches the directory's lock file, and that time is its last use, which decides prune order.
The size is cached in <fingerprint>.size beside the directory. It is measured again only when the directory was used after the size was written.
A new size is recorded only when no job holds the directory, since a held one may still be growing.
A link therefore walks only the directories used since the last prune.

Validation. tests/test_ci_owned_spm_scratch.py (12 tests) passes. The 3 new tests cover:

the holder releasing its lock after its bound;
the size being measured once until the directory is used again;
a held directory not getting a recorded size.

Changelog

none

🤖 Generated with Claude Code

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

Summary by cubic

Follow-up to #14804. Bounds the SwiftPM scratch holder so a runner that dies mid-job can't leave a scratch directory locked until the mini reboots, and caches each directory's size so pruning no longer walks the whole scratch on every link.

The holder now exits after 65 minutes, just past the swift-package-tests job's 60-minute timeout.
link touches the lock file as the directory's last use, and sizes are cached in <fingerprint>.size, re-measured only when a later job used the directory and no job holds it; a time that ties on coarse clocks also re-measures.
New tests cover the holder releasing its lock, size measured once until the directory is used again, and held directories not getting a recorded size.

^{Written for commit 7504716. Summary will update on new commits.}

…ory's size The holder that keeps a job's shared lock on its scratch directory exited only when the runner killed it at job end. A runner that died mid-job left it holding the lock until reboot, so that directory could never be pruned or evicted. It now exits after 65 minutes (signal.alarm), past the swift-package-tests job's 60-minute timeout. tree_stats walked every file of up to 24 GiB of scratch on every link. The lock file, which link now touches, gives a directory's last use, and its size is cached in <fingerprint>.size, measured again only once the directory was used after it and no job holds it. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

cursor · 2026-09-28T12:57:15Z

Bugbot is paused — on-demand spend limit reached

Bugbot uses usage-based billing for this team and has hit its on-demand spend limit.

A team admin can raise the spend limit in the Cursor dashboard, or wait for the next billing cycle to continue.

coderabbitai · 2026-09-28T12:57:18Z

Warning

Review limit reached

Next included review available in 58 seconds.

Check out review usage here.

View limit details

Limit details: You’ve used all 10 included reviews currently available.

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: Repository: manaflow-ai/cmux/.coderabbit.yaml

Review profile: ASSERTIVE

Plan: Advanced

Run ID: 62d9ebc4-8fbb-4ad0-b057-0cc729741c1a

📥 Commits

Reviewing files that changed from the base of the PR and between 31b2619 and 7504716.

📒 Files selected for processing (3)

docs/ci-runners.md
scripts/ci/owned_spm_scratch.py
tests/test_ci_owned_spm_scratch.py

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

github-actions · 2026-09-28T12:57:23Z

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

github-actions · 2026-09-28T12:57:58Z

Dogfood build of 7504716c6b45cece0f32dce4b79e12388fcb216e

cmux DEV pr-15366-7504716c.app

The link opens this exact commit in the cmux dev menu bar app. The build starts on each push and the page waits until it is ready; a newer push replaces it. It signs in against production, so Cloud or backend changes still need a tagged build with a development backend.

blacksmith-sh · 2026-09-28T12:59:51Z

Found 1 test failure on Blacksmith runners:

Failure

Test	View Logs
`test_a_size_is_measured_once_until_the_directory_is_used_again (main.Scratch.test_a` `_size_is_measured_once_until_the_directory_is_used_again)/` `test_a_size_is_measured_once_until_the_directory_is_used_again (main.Scratch.test_a` `_size_is_measured_once_until_the_directory_is_used_again)`	View Logs

^{Need help on this PR? Tag @codesmith-bot with what you need.}

github-actions · 2026-09-28T13:00:12Z

CI failure attribution

CI passes on 7504716c6b (run 36426075832 attempt 1).

Written by scripts/ci/classify_failures.py (ci-failure-attribution.yml); signatures are its SIGNATURES table. A machine verdict is the runner's fault, not this PR's.

…ocks) Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

github-actions · 2026-09-28T13:10:18Z

Merge receipt for 7504716c6b: every check was green at merge (14 verified; 18 skipped by policy). Full suite runs on main after merge.

Resolve the scratch-age overlap using main’s complete implementation and tests from #15366. The cache repair now has no remaining delta from main.

0e298fb ci: wait for the product's canonical root instead of compiling beside it (manaflow-ai#15379) 3088273 ci: UI test runs adopt compile admission's product, skip the re-upload, and report progress (manaflow-ai#15331) b681e7e Keep a pending banner quiet once its pane is focused (manaflow-ai#15357) 03a2f6e Record that cloud_vm_sessions.attachment_count is cumulative (manaflow-ai#15321) 48258b4 fix(iroh-v2): check the team socket cap before opening the session (manaflow-ai#15340) 2638d56 Agent activity reorder follow-ups: group on-top check, search, subtitle (manaflow-ai#15362) 9ed83fd Dogfood journey: record whether a paused Cloud machine is asleep (manaflow-ai#15293) 7171ea8 Add app.tabBarVisibility to hide the pane tab bar when a pane has one tab (manaflow-ai#15294) 8743ec8 test: stop Computer Use onboarding tests waiting out the helper status deadline (manaflow-ai#15329) 6e4f1da ci: drain the snapshot's owned queue by what the machines finished since (manaflow-ai#15374) 9373164 ci: queue a pull request's admission for a root runner when Blacksmith's wait is longer (manaflow-ai#15376) 634a155 test: expect injected pane attention accent (manaflow-ai#15370) cd030e9 Keep a named Cloud machine's prompt name instead of flipping to its slug (manaflow-ai#15288) 24ee0ee Exit 1 when cmux terminal screen wait times out (manaflow-ai#15282) 1b857ac test: cover a live Codex turn owner keeping its turn on SessionStart (manaflow-ai#13588) 56ec600 PR media: prune media of long-closed pull requests (manaflow-ai#15364) 4898cde ci: bound the SwiftPM scratch holder and cache scratch sizes (manaflow-ai#15366) # Conflicts: # .github/workflows/ci-guards.yml # .github/workflows/ci.yml # .github/workflows/test-e2e.yml

ci: re-measure a scratch size whose time ties its last use (coarse cl…

7504716

…ocks) Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

austinywang mentioned this pull request Sep 28, 2026

Allow terminal links to open browser tabs in the same pane #12806

Open

teamleaderleo merged commit 4898cde into main Sep 28, 2026
52 checks passed

teamleaderleo deleted the ci/spm-scratch-holder-bound branch September 28, 2026 13:09

austinywang added a commit that referenced this pull request Sep 28, 2026

Merge main and keep upstream SwiftPM cache policy

1480d0f

Resolve the scratch-age overlap using main’s complete implementation and tests from #15366. The cache repair now has no remaining delta from main.

austinywang mentioned this pull request Sep 28, 2026

Make Cloud Delete Machine optimistic #15190

Merged

3 tasks done

teamleaderleo mentioned this pull request Sep 28, 2026

Keep the main window floor on the animating setFrame path #15368

Merged

github-actions Bot mentioned this pull request Sep 28, 2026

Main full-suite CI is red #15404

Closed

Navigation Menu

Uh oh!

ci: bound the SwiftPM scratch holder and cache scratch sizes - #15366

ci: bound the SwiftPM scratch holder and cache scratch sizes#15366
teamleaderleo merged 2 commits into
mainfrom
ci/spm-scratch-holder-bound

teamleaderleo commented Sep 28, 2026 •

edited by cubic-dev-ai Bot

Loading

Uh oh!

cursor Bot commented Sep 28, 2026

Uh oh!

coderabbitai Bot commented Sep 28, 2026 •

edited

Loading

Review limit reached

Uh oh!

github-actions Bot commented Sep 28, 2026

Uh oh!

github-actions Bot commented Sep 28, 2026 •

edited

Loading

Uh oh!

blacksmith-sh Bot commented Sep 28, 2026

Uh oh!

github-actions Bot commented Sep 28, 2026 •

edited

Loading

Uh oh!

Uh oh!

github-actions Bot commented Sep 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

teamleaderleo commented Sep 28, 2026 • edited by cubic-dev-ai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changelog

Summary by cubic

Uh oh!

cursor Bot commented Sep 28, 2026

Bugbot is paused — on-demand spend limit reached

Uh oh!

coderabbitai Bot commented Sep 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review limit reached

Uh oh!

github-actions Bot commented Sep 28, 2026

Uh oh!

github-actions Bot commented Sep 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

blacksmith-sh Bot commented Sep 28, 2026

Failure

Uh oh!

github-actions Bot commented Sep 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CI failure attribution

Uh oh!

Uh oh!

github-actions Bot commented Sep 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

teamleaderleo commented Sep 28, 2026 •

edited by cubic-dev-ai Bot

Loading

coderabbitai Bot commented Sep 28, 2026 •

edited

Loading

github-actions Bot commented Sep 28, 2026 •

edited

Loading

github-actions Bot commented Sep 28, 2026 •

edited

Loading