Problem
A dropped GitHub push-trigger under heavy trunk contention can leave a fleet lane's seed step waiting for an orchestrate run that never fired (the merge push matched the path filter and was clean, but GitHub did not create the run). Only cascade-example-4env's suite carries a fallback that dispatches orchestrate for the merge SHA when no push-triggered run appears in-window. The other lanes rely on the fleet-level whole-suite retry as a backstop and carry the same latent gap in divergent inline shapes.
Proposed
Normalize the seed fallback across all example-repo scenario suites, ideally onto a shared helper so the logic lives in one place. cascade-example-3env and others use bespoke inline poll loops; cascade-example-rollback-dispatch needs a wait_for_dispatched_run helper added first.
Context
Diagnosed during the v1 fleet-streak work: a 4env lane failed on a dropped trigger, was fixed there, and the fleet-level retry (retry-attempts: 2) currently covers the rare tail elsewhere.
Problem
A dropped GitHub push-trigger under heavy trunk contention can leave a fleet lane's seed step waiting for an orchestrate run that never fired (the merge push matched the path filter and was clean, but GitHub did not create the run). Only
cascade-example-4env's suite carries a fallback that dispatches orchestrate for the merge SHA when no push-triggered run appears in-window. The other lanes rely on the fleet-level whole-suite retry as a backstop and carry the same latent gap in divergent inline shapes.Proposed
Normalize the seed fallback across all example-repo scenario suites, ideally onto a shared helper so the logic lives in one place.
cascade-example-3envand others use bespoke inline poll loops;cascade-example-rollback-dispatchneeds await_for_dispatched_runhelper added first.Context
Diagnosed during the v1 fleet-streak work: a 4env lane failed on a dropped trigger, was fixed there, and the fleet-level retry (
retry-attempts: 2) currently covers the rare tail elsewhere.