Skip to content

docs: update worktree cleanup docs for pipeline liveness gate [doc-updater] - #3106

Merged
jwbron merged 1 commit into
mainfrom
egg/doc-update-worktree-liveness-cleanup
Jun 11, 2026
Merged

docs: update worktree cleanup docs for pipeline liveness gate [doc-updater]#3106
jwbron merged 1 commit into
mainfrom
egg/doc-update-worktree-liveness-cleanup

Conversation

@james-in-a-box

Copy link
Copy Markdown
Contributor

Update documentation to reflect changes from 031f08b / #3087:

  • docs/architecture/git-isolation.md: Crash Recovery section described orphan detection as a simple container-liveness check. Rewrote the steps and pseudocode to show the new pipeline-anchored check, the fail-safe behavior (sweep skipped if orchestrator unreachable), and the invariant that branches are never deleted by orphan sweeps.
  • docs/reference/mcp-deployment-tools.md: prune_stale_worktrees output schema was missing active_pipelines_count. Added it, documented the new 503 response when the orchestrator is unreachable, and noted that worktrees for HITL-parked pipelines are preserved regardless of container liveness.

Triggered by: #3087

Authored-by: egg

Update docs to reflect the new fail-safe behavior introduced in #3087:
orphan sweeps now verify pipeline liveness via the orchestrator before
deleting any worktrees, and skip the sweep entirely if the orchestrator
is unreachable.

- git-isolation.md: rewrite Crash Recovery steps and pseudocode to show
  the pipeline-anchored check, the fail-safe skip on orchestrator
  unavailability, and the no-branch-deletion invariant in orphan sweeps
- mcp-deployment-tools.md: add active_pipelines_count to prune output
  schema, document the 503 response when orchestrator is unreachable,
  and note that HITL-parked pipelines are preserved regardless of
  container liveness

Authored-by: egg
@james-in-a-box

Copy link
Copy Markdown
Contributor Author

egg is investigating the Test check failure...

  • Integration Tests / Integration Tests

@james-in-a-box

Copy link
Copy Markdown
Contributor Author

Check Failure: Integration Tests

What's failing: The "Build containers" step of Integration Tests / Integration Tests failed during docker build of the sandbox image. apt-get update could not fetch the deadsnakes PPA index:

W: Failed to fetch https://ppa.launchpadcontent.net/deadsnakes/ppa/ubuntu/dists/jammy/InRelease  503  Service Unavailable [IP: 185.125.189.187 443]
...
#6 ERROR: process "/bin/sh -c apt-get update && apt-get install -y ... add-apt-repository -y ppa:deadsnakes/ppa ..." did not complete successfully: exit code: 100
make[1]: *** [Makefile:542: build-sandbox] Error 1

Root cause: Transient upstream outage at ppa.launchpadcontent.net (Launchpad). This is an external CDN returning 503, not a problem with the build itself.

Why this is not caused by this PR: The PR is docs-only — it modifies docs/architecture/git-isolation.md and docs/reference/mcp-deployment-tools.md, neither of which is referenced by the sandbox Dockerfile or the integration-test build path.

What needs to be done:

  • Re-run the failed Integration Tests / Integration Tests job. The autofixer cannot invoke gh run rerun (blocked by the gateway), so a human needs to click "Re-run failed jobs" in the Actions UI.
  • If subsequent retries also 503 on the same PPA mirror, consider pinning the apt source or adding a retry loop around add-apt-repository/apt-get update in sandbox/Dockerfile to ride out brief Launchpad outages — but that is a separate hardening change, not something to land on this docs PR.

— Authored by egg

@james-in-a-box

Copy link
Copy Markdown
Contributor Author
Autofix tracking
{"Test/Integration Tests / Integration Tests": 1}

@james-in-a-box

Copy link
Copy Markdown
Contributor Author

egg check fixer completed for Test. CI will re-run to verify. View run logs

— Authored by egg

@jwbron
jwbron merged commit d5b70c3 into main Jun 11, 2026
25 of 28 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant