Bring your towel and join one of the fastest growing data communities. Welcome to our second-generation open source orchestration platform, a completely rethought approach to dataflow automation.

Prefect Community

Hey there - does anyone know the minimum version of cloudpickle required for prefect? I suspect I have a version mismatch, as am getting `unsupported pickle protocol: 5`  as an error on deserializing a flow inside an image that has `cloudpickle==1.2.2` in it. FWIW it serialized fine from inside the same image.

Hi <@U015Z6E2XV5> prefect has the requirements of cloudpickle set to `cloudpickle &gt;=0.6.0` . This is definitely a mismatch issue! Is there a chance you could also check your local version of cloudpickle? Follow up: are you serializing your flow locally on the same python version that your container is running? I feel like I have seen this issue when my container has a newer python version as compared to my local machine :thinking_face:

so that's what I was thinking, but I wasn't seeing that. Trying to sort out some deeper issues with the execution right now (I got that by manually running a kubernetes job with the right permissions) but will dig more.

The way I serialized the flow was to boot into the runtime image using a VSCode devcontainer - I'm not convinced that I've made everything match up perfectly there.