If you are not the administrator of a cluster, but have access to one, this document outlines how to setup a working environment that can be used for optimization.

Note: It is assumed that on this cluster, there is a shared file-system where the software can be installed once and then used by all allocation of resources.

1. Make sure that there is a virtual environment from which you will run Ansible, read the documentation here to set one up.

2. In the cluster-user directory, modify the hosts file to include the cluster head node, the contents should be a single stanza with a single line:

   [neuron-optimizer-worker]
   dns.name.of.head.node
   

3. If the username is different for the cluster from on the host it's being configured from, the user_name in vars.yaml will have to be changed to this new name. One can also set an absolute path in for workspace in vars.yaml to modify the installation path.

4. If the python version available on the cluster isn't at least 2.7, then it is recommended to modify python27_build in vars.yaml to be 'true' such that a local version of Python will be compiled.

Run the installation by issuing:

$ source ./venv-ansible/bin/activate
$ ansible-playbook site.yaml

Once it has run, there should be a ~/workspace directory on the cluster with all the required software as described here under 'Installation Information'.