I'm having an issue when setting @conda_base(pytho...
# dev-metaflow
q
I'm having an issue when setting @conda_base(python='3.9.7') the calling the flow from another computer with local python 3.9.9. The flow is set to use 3.9.7, but the local running python is 3.9.9 The error is
Copy code
Step: start, Error: PackagesNotFoundError: The following packages are not available from current channels:

- python[version='3.9.7,3.9.9']

 Current channels:
....
2
3.9.9 isn't available through conda, anyway, why is concatenating local python if a python version is already set?
s
a problem is that
3.9.9
doesn't exist in channels as you pointed out and
3.9.7
isn't compatible with the current dependencies of Metaflow itself. Can you try e.g.
@conda_base(python='3.8.10')
?
q
context: I'm using prefect to execute metaflow on the cloud (kubernetes) "dynamically" by passing repo url, commit and parameters to a task prior to this message I was using 3.9 prefect image, which contains 3.9.9 hence the error above I just changed to prefect with python 3.8 that actually is 3.8.12 + setting conda_base 3.8.10 Still same error
Copy code
PackagesNotFoundError: The following packages are not available from current channels:

- python[version='3.8.10,3.8.12']
But if I change conda_base to use
3.8.12
it works, maybe because conda base and local python versions are the same To me it seems it's always setting twice python library, 1 from conda_base and another from the local installation It should only use conda_base if specified I think
s
it should, that's odd. Let me debug a bit more to understand what's going on
q
omg I feel so noob 🤦 I was passing the python version as
Copy code
@conda_base(libraries={"python": "3.8.10"})
It should be
Copy code
@conda_base(python="3.8.10")
That's why I had twice the python lib requirement 🤦
all good here 👍
s
ah, awesome!