Channels
#prefect-community
Title
o
Oscar Björhn
08/11/2022, 10:28 AMGot another issue with new deployments, it seems that manifest_path is still required to have a value in the deployment.yaml file. Is this intended?
Copy code
pydantic.error_wrappers.ValidationError: 1 validation error for Deployment
manifest_path
none is not an allowed value (type=type_error.none.not_allowed)✅ 1
a
Anna Geller
08/11/2022, 10:49 AM
Do you use venv? I'd recommend trying with a new fresh conda environment and installing Prefect there
o
Oscar Björhn
08/11/2022, 10:50 AMI do use venv! I re-created it when I upgraded to 2.0.4
Although I don't use conda
a
Anna Geller
08/11/2022, 10:51 AM
again I recommend conda or just not venv, your call to make
I saw one person had similar issue
to troubleshoot and see that venv is at fault by using the old Prefect version, try replacing the default command in YAML with:
Copy code
command:
- python
- -c
- 'import prefect; print(prefect.__version__)'o
Oscar Björhn
08/11/2022, 10:53 AMGiving it a go..
It printed "2.0.4"
I realized this has to do with where I run the agent by the way, that's just a VM that doesn't use venv. Re-created the VM two hours ago and ran pip install prefect==2.0.4, that's about it
I've worked around the issue by pointing it at an old manifest file I had lying around, that I've uploaded to blob storage to make it happy. Probably shouldn't be necessary though. 🙂
🙌 1
d
Deceivious
08/11/2022, 11:30 AMI was facing the same issue. I was using docker image stored on cloud for the agent. I had not updated the docker image with the new version of prefect. The workflows ran successfully after I rebuilt the image and pushed it to dockerhub.
Just posting this in case some one else has the same issue.
🙏 1
o
Oscar Björhn
08/11/2022, 11:35 AMYeah, that explains my issue.. My bad. Thanks Bigya (and Anna)!
🎉 2
a
Anna Geller
08/11/2022, 11:51 AM
Nice work everyone and thanks for sharing both problems and solutions here, very helpful!
4 Views