Hi, question regarding

@retry

on the foreach step and with

--max-workers 1

, metaflow will process 1 child step at a time. if one of the child steps fails, if I run the flow locally, it would retry on the failed child step with the interval, but if I run the flow with argo workflow, it would skip the retry and run other child steps, and then come back to retry on the failed step. Is this expected behavior with argo workflow?

at the moment, there is no invariant on this behavior.

hi @thankful-ambulance-42457 did you have a chance to look at it? cc @hundreds-wire-22547

btw are you able to replicate this within the dev stack?

hmm, we are on 2.14 and a custom metaflow extension, not easy to experiment the dev stack

i don't think there is any dependency between metaflow and dev stack per se