Is anyone seeing issues with prefect cloud rn? I have a task run that says

Trigger was "all_successful" but some of the upstream tasks failed.

but all the upstream tasks were successful. Am I missing something?

Could you give me a screen shot of your schematic?

yes, one sec

That is weird

This is the first time I’ve seen this and have been running this flow almost every day for about 1w I’ll send you the script

I honestly can’t see something immediate. Could you tell me if it happens again?

This thread seems to provide an answer to this - it looks like the List task is the culprit here. @brian could you try to rewrite the List task so that it’s passed to the downstream parcels-silver* task as data dependency rather than as a List collection?

I’m happy to try that @Anna Geller what does it mean to have it be a data dependency?

this thread has an example that shows exactly that

Nice, thanks for sharing that @Anna Geller I’ll give that a whirl. Why would this issue show up just on that one task run though? There are sibling tasks on that same flow run that also accept a List task that completed successfully.

I think it would imply there is maybe something with the content of one of the specific lists but i’m honestly not sure

@Kevin Kho @Anna Geller thank you very much for your help. I’ve made an attempt to convert this particular part of the flow from an implicit List task to a data dependency per the example code y’all shared above. However, this seems to have broken the register step in our CI/CD pipeline since we are using the azure pipelines Bash task to call

prefect register

and have failOnStderr set to

true

(which I don’t really want to change) I now get a warning on stderr that looks like this

/Users/brian.sorahan/Library/Python/3.9/lib/python/site-packages/prefect/storage/base.py:154: UserWarning: Task <Task: Bar> has retry settings but some upstream dependencies do not have result types. See <https://docs.prefect.io/core/concepts/results.html> for more details.
  _healthcheck.result_check(self._flows.values(), quiet=True)  # type: ignore

This is actually me calling register from my laptop, but it’s the same warning we get on azure pipelines. The script I’m trying to register is a small toy script that reproduces the warning. The

list_task

in this script is my workaround for the implicit List task issue we were discussing above, maybe I’m doing that wrong?

Trying it now

I modified the prefect lib locally and have determined that this warning happens because the result of list_task is None

So I looked at it. And I am more confused why it’s not raised for the other upstream tasks that don’t have result. I think the message makes sense and is expected. It will also raise even

list_task

is something like:

@task
def list_task():
    return 2

and then

bar_task = bar(host_name, list_task())

The error makes sense since the retry will grab the output of the last time from its result (especially if Restarting as failed flow run)

I’m confused. The warning seems to complain about the result of list_task not having a type, but it seems like it should have a list type.

It’s not complaining about output types. It’s complaining about Prefect Result types