Hi John,

Thanks for the clarification. I've now installed PBS/torque using the instructions found here:
http://www.discngine.com/blog/2014/6/27/install-torque-on-a-single-node-centos-64

I can correctly submit a job so all is good there. 

However, when I follow the instructions on https://wiki.galaxyproject.org/Admin/Config/Performance/Cluster regarding scrambling the eggs, e.g.:

galaxy_user@galaxy_server% LIBTORQUE_DIR=/path/to/libtorque python scripts/scramble.py -e pbs_python

I get the following:

Traceback (most recent call last):
  File "scripts/scramble.py", line 22, in <module>
    from galaxy.eggs.scramble import ScrambleCrate, ScrambleFailure, EggNotFetchable
ImportError: cannot import name EggNotFetchable

Also, what should my LIBTORQUE_DIR be? I installed this as root in /root/torque-6.0.0.1-1449528029_21cc3d8

Thanks!

On Wed, Jan 20, 2016 at 12:27 PM, John Chilton <jmchilton@gmail.com> wrote:
Even if you just have two servers, I would strongly recommend you
setup a cluster distributed resource manager (DRM) like SLURM, PBS, or
Condor and ensuring there exists a shared file system between Galaxy
and the node running the jobs. You wouldn't even need to use the CLI
job runner - you could just use the DRMAA job runner directly in most
cases if you configured one of these - which is what most people use
to run Galaxy jobs across machines on a cluster.

If you wish to send Galaxy jobs to a single machine without setting up
a DRM or if a shared file system is impossible - you can use Pulsar
(http://pulsar.readthedocs.org/en/latest/) for most kinds of jobs
(some jobs like data source jobs and upload jobs should remain on the
Galxay host in a such a configuration).

-John

On Wed, Jan 20, 2016 at 5:18 PM, D K <danielfortin86@gmail.com> wrote:
> Hi,
>
> I would like to run the Galaxy framework on server A, while performing all
> of the jobs on server B using ssh.
>
> Looking at the documentation here:
> https://wiki.galaxyproject.org/Admin/Config/Performance/Cluster#CLI, this
> seems like it should be possible. However, the documentation states that the
> cli runner requires, at a minimum, two parameters, one for the shell (which
> I'm selecting SecureShell) and a Job plugin. I'm not sure what this should
> be since the ones available are Torque, Slurm, and SlurmTorque and I'm not
> running any of these. Can anyone give me any hints? My current job_conf.xml
> looks like this:
>
> <?xml version="1.0"?>
> <job_conf>
>     <plugins>
> <!--        <plugin id="local" type="runner"
> load="galaxy.jobs.runners.local:LocalJobRunner" workers="4"/>-->
>         <plugin id="cli" type="runner"
> load="galaxy.jobs.runners.cli:ShellJobRunner"/>
>     </plugins>
>     <handlers>
>         <handler id="main"/>
>     </handlers>
>     <destinations default="cli_default">
>         <destination id="cli_default" runner="cli">
>            <param id="shell_plugin">SecureShell</param>
>            <param id="job_plugin">cli</param>
>            <param id="shell_hostname">computehost</param>
>            <param id="shell_username">galaxy</param>
>         </destination>
>     </destinations>
> </job_conf>
>
>
> As an alternative, in a tool xml I tried directly sshing between the
> <command> tags e.g.:
>
>   <command>ssh galaxy@serverB 'sh $__tool_directory__/runTool.sh --verbose >
> $output 2>&amp;1' </command>
>
> This appears to work. Is it possible to "inject" this ssh command into tool
> execution in Galaxy?
>
> I was looking to do this to avoid having to install a scheduler like SGE or
> using Pulsar. Any suggestions would be greatly appreciated.
>
> Thanks!
>
>
> ___________________________________________________________
> 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:
>   https://lists.galaxyproject.org/
>
> To search Galaxy mailing lists use the unified search at:
>   http://galaxyproject.org/search/mailinglists/