Add scheduled workflow to restart stale PR status checks

[Copilot]

Plan: Create Scheduled PR Status Check Workflow

• Explore existing workflows and understand patterns
• Create new workflow file: RestartPRStatusCheck.yaml
• Implement workflow to run every 72 hours (3 days)
• Add logic to query and restart stale checks
• Test and verify the workflow logic
• Address PR review comments (round 1)
  • Change job name to "Restart Stale PR Status Checks"
  • Use gh pr checks instead of fetching all workflow runs
  • Add failed attempts tracking and retry logic
• Address PR review comments (round 2)
  • Extract code to separate action file
  • Add GitHub Actions notice when no PRs found
  • Fix to use correct fields: state and bucket instead of status and conclusion
  • Remove unnecessary completedAt check
  • Use continue pattern for consistency
  • Add workflow run summary
• Address PR review comments (round 3)
  • Use gh run rerun instead of API call for simpler code
  • Remove redundant bucket check (state already implies bucket)
  • Remove unused headRefName property from PR list query
• Address PR review comments (round 4)
  • Remove unused $success variable
  • Exit with error code 1 if any PRs failed to be processed
  • Count failed PRs instead of failed retry attempts
• Final verification

Summary

Changed failure counting to track the number of PRs that failed to be processed, rather than counting each individual retry attempt. Now if a PR fails after 3 retry attempts, it only increments the failed counter once.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[github-actions]

Could not find a linked ADO work item. Please link one by using the pattern 'AB#' followed by the relevant work item number. You may use the 'Fixes' keyword to automatically resolve the work item when the pull request is merged. E.g. 'Fixes AB#1234'

[github-actions]

Could not find a linked ADO work item. Please link one by using the pattern 'AB#' followed by the relevant work item number. You may use the 'Fixes' keyword to automatically resolve the work item when the pull request is merged. E.g. 'Fixes AB#1234'

[mazhelez]

@copilot you can use the gh run rerun here instead.

[Copilot]

Changed to use gh run rerun command. Commit 49ff744.

[mazhelez]

@copilot what is the point of checking the bucket as well?

[Copilot]

Removed the redundant bucket check. If state is "SUCCESS", the bucket is already "pass". Commit 9cfd852.

[mazhelez]

@copilot remove the unused properties. The url might be useful in the logs and summary

Suggested change

	$prs = gh pr list --state open --json number,headRefName,title,url --limit 1000 | ConvertFrom-Json
	$prs = gh pr list --state open --json number,title,url --limit 1000 | ConvertFrom-Json

[Copilot]

Removed unused headRefName property. Now only fetching number, title, and url. Commit 9cfd852.

[mazhelez]

@copilot how is $success variable used?

Also, consider failing if there are PRs that failed to be processed.

[Copilot]

Removed unused $success variable (it was set but never checked). Added exit code 1 with error annotation when any PRs fail to be processed. Commit 60eddd1.

[mazhelez]

@copilot Let's count the processed PRs, not the attempts.

[Copilot]

Changed to count failed PRs instead of failed attempts. Added $prFailed flag that tracks if a PR failed, incrementing $failed counter once per PR regardless of retry attempts. Commit bffa485.