Hi there, got a question about the following: • I've got a workflow with a task that sometimes may fail due to a timeout or another reason. Is it possible to "ignore" the failure somehow or make the task optional that regardless if it succeeds or fails I would like to continue the workflow run (and if necessary replace the expected output of the failed task with some dummy default output)

is this handleable in the user code?

Given that it mostly fails due to timeouts (set on the task in the workflow definition), I don't think so. UNless I can gracefully shutdown the process by intercepting a signal sent by flyte

i don’t think so… there is no mechanism by which we allow errors to not be a concern. there’s the workflow failure policy that you can configure, but that will still fail the workflow. alternatively (and I haven’t tried this) but you can attempt to create a map task of one element with a min success ratio of 0.

Any pointers as to what signal to expect? Or is it just a `KeyboardInterrrupt`/`SigIn` /`sigKill`?

i forget what the signal is, i can ask someone internally who might know more.

In this case, it was because the task was calling a service that had internal errors and I had setup retries to keep repeating the request until a 200 status would come (which would never happen). Fixed that, so all good now

let me dig

But, just for sake of fault tolerance and best practices, still interested in the signal. Thank you!