Not sure why this is happening, and I don't think this behavior occured with previous Nebari versions (the ESIP deployment at v2023.10.1 doesn't have this issue)
Expected behavior
Expected behaviour is if only conda-forge channel is specified, packages will be installed from conda-forge, not pip.
OS and architecture in which you are running Nebari
Describe the bug
I upgraded to Nebari version 2023.11.1 and built a new environment (called
pangeo
) in the global namespace using conda-store.I checked versions of
bokeh
andholoviz
after the environment built and was surprised to find a mix of packages from conda-store and pip:This is despite my specifying only the
conda-forge
channel in my environment:Not sure why this is happening, and I don't think this behavior occured with previous Nebari versions (the ESIP deployment at v2023.10.1 doesn't have this issue)
Expected behavior
Expected behaviour is if only
conda-forge
channel is specified, packages will be installed fromconda-forge
, not pip.OS and architecture in which you are running Nebari
AWS
How to Reproduce the problem?
Create an environment with the above yaml.
Command output
No response
Versions and dependencies used.
Kubenetes 1.26 Nebari 2023.11.1
Compute environment
AWS
Integrations
No response
Anything else?
No response