Changing Target Machine

All of our examples so far have been run locally. Its time to run something on a HPC! One of the features of Ensemble Toolkit is that you can submit tasks on another machine remotely from your local machine. But this has some requirements, you need to have passwordless ssh access to the target machine. If you don’t have such access, we discuss the setup here. You also need to confirm that RP and Ensemble Toolkit are supported on this machine. A list of supported machines and how to get support for new machines is discussed here.

Note

The reader is assumed to be familiar with the PST Model and to have read through the Introduction of Ensemble Toolkit.

Note

This chapter assumes that you have successfully installed Ensemble Toolkit, if not see Installation.

Once you have passwordless access to another machine, switching from one target machine to another is quite simple. We simply re-describe the resource dictionary that is used to create the Resource Manager. For example, in order to run on the ACCESS Stampede cluster, we describe the resource dictionary as follows:

    'project': 'unc100',
    'schema': 'gsissh'

}

# Assign resource request description to the Application Manager
appman.resource_desc = res_dict

# Run the Application Manager

You can download the complete code discussed in this section here or find it in your virtualenv under share/radical.entk/user_guide/scripts.

python change_target.py

Let’s take a look at the complete code in the example. You can generate a more verbose output by setting the environment variable RADICAL_ENTK_VERBOSE=DEBUG.

A look at the complete code in this section:

#!/usr/bin/env python

from radical.entk import Pipeline, Stage, Task, AppManager
import os

# ------------------------------------------------------------------------------
# Set default verbosity
if os.environ.get('RADICAL_ENTK_VERBOSE') is None:
    os.environ['RADICAL_ENTK_REPORT'] = 'True'


if __name__ == '__main__':

    # Create a Pipeline object
    p = Pipeline()

    # Create a Stage object
    s1 = Stage()

    # Create a Task object which creates a file named 'output.txt' of size 1 MB
    t1 = Task()
    t1.executable = '/bin/bash'
    t1.arguments = ['-l', '-c', 'base64 /dev/urandom | head -c 1000000 > output.txt']

    # Add the Task to the Stage
    s1.add_tasks(t1)

    # Add Stage to the Pipeline
    p.add_stages(s1)

    # Create another Stage object
    s2 = Stage()
    s2.name = 'Stage 2'

    # Create a Task object
    t2 = Task()
    t2.executable = '/bin/bash'
    t2.arguments = ['-l', '-c', 'grep -o . output.txt | sort | uniq -c > ccount.txt']
    # Copy data from the task in the first stage to the current task's location
    t2.copy_input_data = ['$Pipline_%s_Stage_%s_Task_%s/output.txt' % (p.name,
        s1.name, t1.name)]
    # Download the output of the current task to the current location
    t2.download_output_data = ['ccount.txt']

    # Add the Task to the Stage
    s2.add_tasks(t2)

    # Add Stage to the Pipeline
    p.add_stages(s2)

   # Create Application Manager
    appman = AppManager()

    # Assign the workflow as a set or list of Pipelines to the Application Manager
    appman.workflow = set([p])

    # Create a dictionary to describe our resource request for XSEDE Stampede
    res_dict = {

        'resource': 'xsede.comet',
        'walltime': 10,
        'cpus': 16,
        'project': 'unc100',
        'schema': 'gsissh'

    }

    # Assign resource request description to the Application Manager
    appman.resource_desc = res_dict

    # Run the Application Manager
    appman.run()