Tool dependencies resolved by conda but tool will use default galaxy virtualenv to run

1

Hello,

I have a series of non-tool-shed tools I would like to install on my Galaxy instance.

macros.xml includes following requirements for all of the tools

requirement type=“package” version=“2.7” - python - requirement
requirement type=“package” version=“1.65” - biopython - requirement
requirement type=“package” version=“2.12.1” - requests - requirement

All of these are resolved by conda in my instance, but when I run a tool, it uses default galaxy virtualenv python executable instead of mulled conda environment causing errors. How should I let the job run inside conda environment?

2

When you execute such a job, do you see the Conda environment being resolved in the Galaxy logs ?
In the galaxy.yml file, did you set preserve_python_environment to a non-default value ?
Does this happen for all tools or just python tools ?

3

Thanks for your response. This issue seems to be resolved, and it seems like I caused the problem. I was running run.sh inside Ubuntu Screen instance so Galaxy can keep running when I close ssh. And Screen apparently initiated another bash shell which seems to have caused this issue. I now run inside tmux and it seems to have fixed the issue.