microsoft / microsoft/vscode

Agent Merge: retry timed-out CI jobs once before giving up

Open
#334,663 0 comments 0 reactions 1 assignee Claimed by @benibenj View on GitHub
agents-window
Dominant language
TypeScript
Stars
193k
Forks
42.4k
PR merge metrics
PR metrics pending

Description

Testing #334491

## Description
Agent Merge should automatically retry a failed required CI check at least once when the failure is a timeout that may be caused by transient infrastructure problems, rather than immediately giving up.

## Steps to reproduce
1. Enable Agent Merge with permission to fix failed required CI checks.
2. Have a required CI job fail due to a timeout.
3. Observe how Agent Merge handles the failure.

## Expected
- Detect a timeout failure that may be infrastructure-related.
- Automatically rerun the failed job once before giving up, without requiring a code change.
- If the retry also fails, stop automatic retries and surface the remaining failure for investigation.
- Track the retry across Agent Merge turns so the same failure does not trigger an unlimited retry loop.

A timeout is not necessarily a code regression; one bounded retry can recover from a transient runner or infrastructure issue. A timeout alone should not be treated as proof that the failure is infrastructure-related.

## Actual / requested improvement
Agent Merge can give up on a timeout failure without first attempting a retry. Provide a bounded automatic retry before reporting that it cannot proceed.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.