Hey
@Mark McDonald, I think if you don’t want to use FAIL or SKIP, then you would need to use a state handler to hit the GraphQL API to cancel the run. This is a best effort thing because it’s hard to stop processing that is happening on other compute (think of stuff sent to a Dask cluster). See
this recent issue.
You can also maybe use a terminal state handler at the end to check if the last task had a TRIGGERFAILED or SKIPPED, and then return a cancelled state from that