Configuring Jobs

From Daya Bay

Jump to: navigation, search
Offline Documentation: [Offline Category] [FAQ] [Howto] [Reference] [Manual]


Offline Documentation
This article is part
of the Offline Documentation
Other pages...

Offline Category
FAQ
How Tos
Reference
Manual
Getting Started
Software Installation

See also...

General Help on this Wiki


Note: this needs cleaning up.

Here is a quick introduction to configuring your jobs in the offline software. For a detailed presentation, see Brett's slides from the Dec08 tutorial DocDB-2941.

Using Gaudi

At the center of each job is Gaudi and you often do not "see" this center.

You interact with Gaudi through Python-based job configuration code and by writing algorithms in either Python or C++.

You then run a job through nuwa.py

Using nuwa.py

Essentially all possible jobs are run through a single command line executable called nuwa.py. It provides commonly needed options and can run your "Job Option Modules". See the Nuwa.py topic for details.

Job Option Modules

To make a job unique you must a Job Option Module (JOM). Adding modules to a job is done like:

nuwa.py -m MyModule

Modules can take their own command line arguments, in which case be sure to place things in quotes:

nuwa.py -m 'MyModule [options] [arguments]' ...

See the Job Option Module topic for details.

Using GaudiAlgo

Using this approach, nuwa.py plus a Module, you can only run the algorithms that are C++ and have been compiled into dybgaudi. What if you want to do a simple job, like filling histograms or making a simple calculation? Gaudi has a sneaky feature that you can define an Algorithm in python and pass it into Gaudi. This is called a GaudiAlgo. So in your Module, you can write a python class based on GaudiAlgo:

class YourNewAlg(GaudiAlgo):
      /// define class functions initialize(), execute(), finalize() here ///

Because of the details of Gaudi, you can only add your python algorithm just before starts 'running'. To do this, you must define a 'run(app)' function in your Module:

def run(app):
    
    Configure and add an algorithm to job
    
    yourAlg = YourNewAlg("YourNewAlg")
    app.addAlgorithm( yourAlg )

Then you can run the Gaudi job:

% nuwa.py -n 100 YourGaudiAlgoModule

Using Configure() classes

If you have a C++ Algorithm and the configuration is complicated, you can put all the configuration into a python class and just call the Configure() class. This class can be put in a separate file ('YourAlgConfiguration.py') and called using the python 'import' function.

In YourAlgConfiguration.py file:

class Configure()
    def __init__():
         "Set up complicated configuration here"
         ...
         ...
         ...

In YourModule.py file:

import YourAlgConfiguration
def configure():
     "Call your helper class to set configuration"
     myConfigure = YourAlgConfiguration.Configure()

Then you can run it:

% nuwa.py -n 100 YourModule
Offline Software Documentation: [Offline Categories] [FAQ] [Offline Manual Category]
Personal tools