nodejs / nodejs/node-core-utils

pr_checker: ignore superseded GitHub Actions check suites when a head SHA is reused

Open
#1,151 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
JavaScript
Stars
314
Forks
137
Avg merge
2d 8h
Merged PRs (30d)
24

Description

PRChecker.checkGitHubCI() can incorrectly report a PR as having failed GitHub CI when its current head commit has both:

  • an older cancelled/failed GitHub Actions run; and
  • a newer successful run of the same workflow.

This can happen when a PR branch is force-pushed away from a commit and later force-pushed back to exactly the same SHA. GitHub
keeps both sets of check suites attached to that commit.

This occurred on nodejs/node#64803, https://github.com/nodejs/node/pull/64601 whose current head SHA is:

ee1ca501264411a3e993d4ed487f833fb0ffa841
a23d7a9a84a190e3804f9b2fb200feb833f3114a

Timeline:

  1. An Actions run was started for this SHA:
    https://github.com/nodejs/node/actions/runs/30359390795
  2. The PR branch was updated, causing that run to be cancelled by the workflow's cancel-in-progress configuration.
  3. The branch was subsequently force-pushed back to the same SHA:
    https://github.com/nodejs/node/pull/64803#event-28598521231
  4. A new run for the same SHA completed successfully:
    https://github.com/nodejs/node/actions/runs/30360481950
  5. The latest Jenkins CI also completed successfully:
    https://ci.nodejs.org/job/node-test-pull-request/75495/

However, the commit queue reported one failed and 15 cancelled GitHub Actions jobs from the older, superseded runs:

https://github.com/nodejs/node/pull/64803#issuecomment-5226872042

The commit checks page contains both the old cancelled suites and the newer successful suites:

https://github.com/nodejs/node/commit/ee1ca501264411a3e993d4ed487f833fb0ffa841/checks

The reported macOS failure was also produced while the old workflow was being cancelled, rather than by a test failure:

https://github.com/nodejs/node/actions/runs/30359390795/job/90275679554

The GraphQL query retrieves up to 100 check suites associated with the current commit:

https://github.com/nodejs/node-core-utils/blob/v7.0.1/lib/queries/PRCommits.gql#L28-L44

checkGitHubCI() then iterates over every GitHub Actions suite and treats any non-successful historical suite as a current
failure:

https://github.com/nodejs/node-core-utils/blob/v7.0.1/lib/pr_checker.js#L461-L515

It does not identify whether an older suite has been superseded by a newer run of the same workflow.

As a result, an obsolete cancelled or failed run can permanently prevent the commit queue from accepting that SHA, even though
all current CI runs succeeded.

For each GitHub Actions workflow, only the latest applicable run for the current head SHA should determine the CI result. Older
runs superseded by a newer run of the same workflow should be ignored.

Refs: https://github.com/nodejs/node/pull/64601#issuecomment-5226071594
Refs: https://github.com/nodejs/node/pull/64803#issuecomment-5226872042

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with lib/pr_checker.js: checkGitHubCI() iterates over the suites returned by lib/queries/PRCommits.gql. Use the linked nodejs/node examples to understand the duplicated runs for one head SHA, then verify that only the latest applicable run for each workflow affects the result and older superseded runs do not block acceptance.

Written by the indexing model from the issue text.

Assessment

Tech stack
github-actions, javascript
Domain
ci-cd, tooling
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
68/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.