Thanks guys,

To Ray, I didn't mean the way to deploy galaxy application  itself (as a roll). I asked how people setup a submission host in where galaxy will be install. Definitely Rocks frontend is one option.

I guess Shantanu's idea is closer to mine, deploy a galaxy appliance. I could think of such appliance is extended from the default "Login Appliance" (which is merely just a submission host for the users to submit jobs etc).

I am also trying to  setup an standalone host then register it in the Rocks SGE cluster, Shantanu which SGE distribution you used to install on your galaxy VM (I guess it's Centos 5 host)?

Cheers,
Derrick

On Fri, Feb 3, 2012 at 7:09 AM, Shantanu Pavgi <pavgi@uab.edu> wrote:

On Feb 1, 2012, at 9:13 PM, Derrick LIN wrote:

> Hi guys,
>
> Is there anyone has actually setup galaxy on a Rocks based cluster?
>
> I know that galaxy should be hosted on a submission host so that it can submit jobs to the cluster.
>
> Rocks provides preconfigured DRM such as SGE, so I start thinking if galaxy host should be deployed and managed by Rocks (e.g. as a Login Appliance in Rocks' term).
>
> Sorry if this is not a pure galaxy issue, but I guess people in here will be more appropriate to answer this.
>
> Regards,
> Derrick
> ___________________________________________________________
> Please keep all replies on the list by using "reply all"
> in your mail client.  To manage your subscriptions to this
> and other Galaxy lists, please use the interface at:
>
>  http://lists.bx.psu.edu/


We have a similar setup. We had thought about creating galaxy appliance using Rocks but we haven't done so yet.  Instead our main galaxy instance is running as a VM which is distinct from Rocks-SGE cluster. The galaxy VM is registered as an SGE submit host with the SGE master host which is part of the Rocks cluster. The cluster and the galaxy VM have a shared storage so our install uses unified method as described on the galaxy wiki. The galaxy environment (PATH, LD_LIBRARY_PATH, PYTHONPATH, TEMP_DIR etc..) is managed using environment module files which we auto-load for 'galaxy' user using .bashrc file.

In addition to the main galaxy VM instance our cluster users can setup their personal galaxy instances in their home directory for tool development or any other work that's not currently supported in the main instance. And users/developers can run galaxy tool wrappers directly from command-line if needed.

I guess I have added too many details that  you hadn't actually asked about. But just wanted to share an overview of our galaxy install. Hope this is helpful.

--
Shantanu