fix: recover fork PR re-runs when GitHub omits PR metadata by chmouel · Pull Request #2655 · tektoncd/pipelines-as-code

chmouel · 2026-04-08T09:02:29Z

When GitHub receives a Re-run request for a Pipelines as Code check tied to a pull request from a fork, the webhook payload can omit the usual pull request metadata. PAC could already recover by looking up pull requests for the head SHA, but fork PRs still failed when GitHub's commits/{sha}/pulls API returned no matches.

This change makes rerequest handling more resilient and keeps the selection rules safe:

check_run rerequests now honor pull request data attached directly to the event when GitHub provides it
if the commit lookup API returns no open PR, PAC falls back to listing open pull requests and matching by head SHA so fork PR rerequests can still resolve
the SHA fallback now preserves the existing ambiguity guard and fails if more than one open PR shares the same head SHA instead of picking one arbitrarily
tests now cover the direct-event path, fork fallback success, missing-PR failures, and the ambiguous fallback regression

📝 Description of the Change

Improve GitHub rerequest PR resolution so fork pull requests can be recovered safely even when GitHub omits PR metadata or the commit lookup API returns no match.

🔗 Linked GitHub Issue

Not linked.

🧪 Testing Strategy

Ran:

go test ./pkg/provider/github

🤖 AI Assistance

I have not used any AI assistance for this PR.
I have used AI assistance for this PR.

AI assistance was used to help draft and refine the code change, tests, and PR description. The resulting changes were reviewed and validated with the package test run above.

✅ Submitter Checklist

codecov-commenter · 2026-04-08T09:05:17Z

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

❌ Patch coverage is 96.00000% with 3 lines in your changes missing coverage. Please review.
✅ Project coverage is 58.94%. Comparing base (c9be9d6) to head (6ecab48).

Files with missing lines	Patch %	Lines
pkg/provider/github/parse_payload.go	96.00%	2 Missing and 1 partial ⚠️
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2655      +/-   ##
==========================================
+ Coverage   58.85%   58.94%   +0.08%     
==========================================
  Files         204      204              
  Lines       20149    20193      +44     
==========================================
+ Hits        11859    11903      +44     
  Misses       7525     7525              
  Partials      765      765

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

gemini-code-assist

Code Review

This pull request introduces a fallback mechanism to resolve pull requests by SHA when the standard commit API fails, which is particularly useful for fork PRs. It also adds logic to extract pull request information directly from the check_run event. Feedback focuses on ensuring consistency and handling ambiguity when multiple pull requests might match a single SHA or check run, suggesting the use of existing helper functions to validate that only a single open pull request is processed.

chmouel · 2026-04-08T11:07:12Z

tested on dogfooding as working

zakisk · 2026-04-10T14:17:28Z

+
+	// ListPullRequestsWithCommit may return no matches for fork PR commits.
+	v.Logger.Infof("No PR found via commits API for SHA %s, falling back to open PR listing", runevent.SHA)
+	return v.findOpenPullRequestBySHA(ctx, runevent.Organization, runevent.Repository, runevent.SHA)


can we call the findOpenPullRequestBySHA only if its fork because if its not fork and there is no matching PRs as well then it would be just redundant API calls! (hint: event.pull_request.head.fork)

zakisk · 2026-04-10T14:18:37Z

 	// If we don't have a pull_request in this it probably mean a push
 	if len(event.GetCheckRun().GetCheckSuite().PullRequests) == 0 {
+		if len(event.GetCheckRun().PullRequests) > 0 {
+			runevent.PullRequestNumber = event.GetCheckRun().PullRequests[0].GetNumber()


is it certain here that first PR would be the exact PR where rerun is made?

there is no guarantee enough to keep using PullRequests[0] blindly, so I added a small guard for this path.

check_run.pull_requests now fails fast if GitHub gives us more than one associated PR in the webhook payload instead of guessing.

check_run.pull_requests now fails fast if GitHub gives us more than one associated PR in the webhook payload instead of guessing.

you're going to add or you did it? I don't see anything?

When someone clicks Re-run on a Pipelines as Code check for a pull request coming from a fork, GitHub can leave out the usual pull request details in the webhook payload. That left PAC with no obvious way to tell which pull request the check belonged to, so the re-run stopped with an error instead of starting again. This change teaches PAC to keep looking in a more practical way. If the first GitHub API says it cannot find the pull request for the commit, PAC now lists open pull requests in the repository and matches the one whose head commit SHA is the same. In simple terms: when GitHub does not hand us the answer directly, we now look through the open pull requests and find the one built from that exact commit. The change also checks pull request data attached directly to the check_run event before falling back to SHA-based lookup, and the tests now cover both the new fork pull request path and the empty-result error cases. Signed-off-by: Chmouel Boudjnah <chmouel@redhat.com>

zakisk · 2026-04-15T06:32:53Z

+	if len(checkSuite.PullRequests) > 0 {
+		runevent.PullRequestNumber = checkSuite.PullRequests[0].GetNumber()
+		runevent.TriggerTarget = triggertype.PullRequest
+		v.Logger.Infof("Recheck of PR %s/%s#%d has been requested", runevent.Organization, runevent.Repository, runevent.PullRequestNumber)
+		return v.getPullRequest(ctx, runevent)
+	}


Suggested change

if len(checkSuite.PullRequests) > 0 {

runevent.PullRequestNumber = checkSuite.PullRequests[0].GetNumber()

runevent.TriggerTarget = triggertype.PullRequest

v.Logger.Infof("Recheck of PR %s/%s#%d has been requested", runevent.Organization, runevent.Repository, runevent.PullRequestNumber)

return v.getPullRequest(ctx, runevent)

}

you can delete this as you're already doing this on line 557 below

gemini-code-assist bot reviewed Apr 8, 2026

View reviewed changes

Comment thread pkg/provider/github/parse_payload.go

Comment thread pkg/provider/github/parse_payload.go Outdated

chmouel marked this pull request as draft April 8, 2026 09:11

chmouel force-pushed the fix-check-run-rerequest-fork-pr branch from 3282631 to 212adf3 Compare April 8, 2026 09:17

chmouel marked this pull request as ready for review April 8, 2026 11:07

chmouel force-pushed the fix-check-run-rerequest-fork-pr branch from 5f94900 to 8a5873e Compare April 8, 2026 14:06

zakisk reviewed Apr 10, 2026

View reviewed changes

chmouel marked this pull request as draft April 14, 2026 13:25

chmouel force-pushed the fix-check-run-rerequest-fork-pr branch from 8a5873e to c6a4c49 Compare April 14, 2026 15:30

chmouel force-pushed the fix-check-run-rerequest-fork-pr branch from c6a4c49 to 6ecab48 Compare April 14, 2026 15:30

zakisk reviewed Apr 15, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix: recover fork PR re-runs when GitHub omits PR metadata#2655

fix: recover fork PR re-runs when GitHub omits PR metadata#2655
chmouel wants to merge 1 commit intotektoncd:mainfrom
chmouel:fix-check-run-rerequest-fork-pr

chmouel commented Apr 8, 2026 •

edited

Loading

Uh oh!

codecov-commenter commented Apr 8, 2026 •

edited

Loading

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

Uh oh!

Uh oh!

chmouel commented Apr 8, 2026

Uh oh!

zakisk Apr 10, 2026

Uh oh!

zakisk Apr 10, 2026

Uh oh!

chmouel Apr 14, 2026 •

edited

Loading

Uh oh!

zakisk Apr 14, 2026

Uh oh!

zakisk Apr 15, 2026

Uh oh!

zakisk Apr 15, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

chmouel commented Apr 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

📝 Description of the Change

🔗 Linked GitHub Issue

🧪 Testing Strategy

🤖 AI Assistance

✅ Submitter Checklist

Uh oh!

codecov-commenter commented Apr 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

Uh oh!

chmouel commented Apr 8, 2026

Uh oh!

zakisk Apr 10, 2026

Choose a reason for hiding this comment

Uh oh!

zakisk Apr 10, 2026

Choose a reason for hiding this comment

Uh oh!

chmouel Apr 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zakisk Apr 14, 2026

Choose a reason for hiding this comment

Uh oh!

zakisk Apr 15, 2026

Choose a reason for hiding this comment

Uh oh!

zakisk Apr 15, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

chmouel commented Apr 8, 2026 •

edited

Loading

codecov-commenter commented Apr 8, 2026 •

edited

Loading

chmouel Apr 14, 2026 •

edited

Loading