The cloud-native orchestrator for the whole development lifecycle, with integrated lineage and observability.

dagster

maybe not exactly what you're looking for, but have you considered using the underlying graphs for the jobs to make a job with nested graphs, so then you can just execute it as a single job with order-based dependencies between the graphs?

I don't think there's much of a way to interact with the queuing coordinator to evict / cancel queued jobs via dependency failures

Thanks for confirming, we’ll handle it differently and set up run_status_sensors to trigger runs sequentially I think.