Can a failed DAG task force all running tasks to stop immediately?

[csm-kb]

Question above! Core context:

• My application contains two pods that talk to one another one-way through a messaging interface (think sensor information to a processing service).
• I have written an Argo workflow that performs an end-to-end test against specified versions of both of these pods. They spawn up, talk to each other, and then resolve successfully.
• However, if an error causes the sensor pod to fail abruptly, the processing service runs as normal, waiting for a termination message that will never arrive...

ℹ I'm looking to stop the processing DAG task when the other branch fails, so the workflow can resolve onExit and fail gracefully instead of being left uncaught (except via e.g. long CI times).

• Is this currently possible with Argo?

From what I've read, I don't think so -- FailFast only skips scheduling of further DAG tasks, and a deep dive of issues and discussions didn't reveal anyone asking this exact "stop the whole DAG immediately if one task fails" question.

I am aware that I could incorporate some odd logic into my pods or write a sidecar to handle this testing edge-case -- but this could be a beneficial Argo DAG config setting, if this isn't currently possible and there are other use cases that support it!

[Reversaidx]

have the same question