ci-workflow-guide

CI test partitioning splits staged test execution across multiple runners to enable fast-fail PR gating. This architecture surfaces root causes quickly and minimizes wait times by running base, kernel, and advanced test stages in parallel.

You configure CI workflows by defining base, kernel, and advanced test stages with appropriate gating and fast-fail machinery. This staged test execution ensures reliable CI feedback through cross-stage health checks and automated progression.

SGLang CI pipelines fail to fast-fail when gating and partitioning are misaligned across test stages. Correctly orchestrating the base, kernel, and advanced stages ensures fast-fail machinery surfaces root causes quickly without waiting for all tests to complete.

You need to align CI gating and fast-fail machinery when a PR introduces a new test stage to the orchestration pipeline. Proper alignment minimizes wait times and ensures cross-stage health checks provide reliable CI feedback.

CI workflow orchestration requires configuring test runners to manage stage progression and health checks for test partitioning. Runners must support the base, kernel, and advanced stages to properly execute gating and fast-fail logic.

The best way to debug CI failures in scheduled runs is analyzing the staged test orchestration architecture. Reviewing the gating, fast-fail machinery, and cross-stage health checks helps isolate whether failures originate in base, kernel, or advanced test stages.