You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In each failed case, prefect was happy to let me deploy, but then failed when actually running the flow. This feels like it could be validated upfront to dramatically reduce my cycle time as a developer.
Version info
❯ uv run prefect version
Version: 3.1.15
API version: 0.8.4
Python version: 3.11.9
Git commit: 3ac3d548
Built: Thu, Jan 30, 2025 11:31 AM
OS/Arch: darwin/arm64
Profile: ephemeral
Server type: cloud
Pydantic version: 2.10.4
Integrations:
prefect-github: 0.3.1
Additional context
No response
The text was updated successfully, but these errors were encountered:
just as another real-world use-case, i just accidentally did
- prefect.deployment.steps.set_working_directory:
instead of - prefect.deployments.steps.set_working_directory: (missing s) and again it deployed happily and then I got the confusing error message on running my flow ModuleNotFoundError: No module named 'prefect.deployment'
we could do validation of build / push / pull steps up front as we could load the associated function (all steps are fully qualified function names) and check the provided values against their signatures
as for (what appears to be) your case though, is it possible you don't need set_working_directory here? if you're just trying to clone a remote repo, the git_clone step alone should suffice. set_working_directory is typically only needed if you're baking code into an image or if you have a process worker looking for source in a bunch of different places on the same machine
thanks @zzstoatzz -- since we are chatting about the actual set_working_directory issue in slack, i'll leave that out from here.
Seems like you potentially have a matching issue in #16717, so if you think that encapsulates the issue better than this one then feel free to close :)
Bug summary
I was working through this guide https://docs.prefect.io/v3/deploy/infrastructure-concepts/prefect-yaml#the-pull-action and was struggling to set the
set_working_directory
step because it's not actually specified how to use it. I triedas well as
(which is what prefect docs AI said, but didn't make sense to me)
as well as a few others, until I searched for the usage of it in the repo to find the right syntax.
In each failed case, prefect was happy to let me deploy, but then failed when actually running the flow. This feels like it could be validated upfront to dramatically reduce my cycle time as a developer.
Version info
Additional context
No response
The text was updated successfully, but these errors were encountered: