By the way the comment is written, conda should be used as a fallback, but instead it is called unconditionally, after installation with mamba. This, in environments that are large, can lead to unacceptably long (days) times to install packages, given conda's atrociously slow solver. And even if it did not, installing packages twice makes no sense.
The issue lies in
https://github.com/chapmanb/cloudbiolinux/blob/924029a5614bfb730ae8e980b0b879a87cf42ba8/cloudbio/package/conda.py#L67
and
https://github.com/chapmanb/cloudbiolinux/blob/924029a5614bfb730ae8e980b0b879a87cf42ba8/cloudbio/package/conda.py#L71
By the way the comment is written, conda should be used as a fallback, but instead it is called unconditionally, after installation with mamba. This, in environments that are large, can lead to unacceptably long (days) times to install packages, given conda's atrociously slow solver. And even if it did not, installing packages twice makes no sense.
Ref https://github.com/bcbio/bcbio-nextgen/issues/3395