ci-workflow-guide

Debug skipped GitHub Actions jobs in staged CI workflows by analyzing cross-stage gating conditions, concurrency limits, and fast-fail behavior. Verify stage dependencies, runner assignments, and continue-on-error settings to isolate the root cause of the skipped execution.

Matrix partitioning for GPU test suites uses Longest Processing Time (LPT) logic to distribute tests across runners. It balances execution load by registering suites, assigning matrix counts, and configuring retry classifications to optimize continuous integration throughput.

Add new test stages to CI pipelines by mapping workflow dependencies, updating matrix counts, and registering test suites. Ensure cross-stage gating logic, health checks, and concurrency settings are updated to maintain reliable stage ordering and prevent pipeline breaks.

CI workflow timeouts during GPU testing stages often stem from unbalanced matrix partitioning, inadequate runner assignments, or missing health checks. Investigate test dispatch configurations, retry policies, and fast-fail behavior to identify execution bottlenecks.

Yes, you can configure continue-on-error settings for flaky test retries in GitHub Actions. The workflow orchestration logic documents retry classification rules, allowing specific test suites to fail without triggering fast-fail behavior or blocking downstream CI stages.

Fast-fail behavior in staged continuous integration workflows is an execution mode that cancels subsequent stages when a critical job fails. It prevents unnecessary resource consumption by terminating parallel matrix jobs and skipping downstream test suites.