On Tue, Oct 2, 2012 at 2:15 PM, Nate Coraor <nate@bx.psu.edu> wrote:
On Oct 2, 2012, at 12:45 PM, greg wrote:
Hi guys,
I'm following the instructions here to install Galaxy on our SGE cluster. http://wiki.g2.bx.psu.edu/Admin/Config/Performance/Cluster (I'm aiming for the unified install)
Here are a couple of questions I'm hoping someone could clear up for me:
How exactly do I use virtual env to make Galaxy install all the Python libraries there instead of on the system? How will Galaxy know to use the virtualenv to run and to use it for grid jobs?
Hi Greg,
You should start at the top level documentation for production servers:
http://wiki.g2.bx.psu.edu/Admin/Config/Performance/Production%20Server
This includes instructions on how to set up a virtualenv. You can make this the default by setting the $PATH in the Galaxy user's ~/.bash_profile to include the virtualenv's bin directory (at the beginning of $PATH).
This makes sense, but I ultimately want to run the jobs as the logged in user as per this page: http://wiki.g2.bx.psu.edu/Admin/Config/Performance/Cluster#Submitting_Jobs_a... It wouldn't be feasible to add the virtualenv to every users' path. I'm hoping there's some kind of other way? Thanks again, Greg