View source for TUESDAY

'''TUESDAY'S OBJECTIVES'''
* Run your Athena job on the grid (this will take your whole day). 
''' ATHENA ON THE GRID'''
* Write a batch script which runs your Athena job on the local machine.
* Write the corresponding xrsl file. Submit it to the grid (like described for Monday).
''' Example which compiles on remote cluster :'''

[shaug@lheppc44] > cat test.sh 
 #/bin/bash
 # /terabig/shaug/atlas
 # tar cvf myAnalysis.tar 14.1.0.2/
 pwd

 ls -l 
 env
 export CMTPATH=$HOME:$CMTPATH
 cd 14.1.0.2/PhysicsAnalysis/AnalysisCommon/UserAnalysisUtils/cmt
 source setup.sh
 cmt config
 gmake
 cd
 cd 14.1.0.2/PhysicsAnalysis/AnalysisCommon/UserAnalysis/cmt
 cmt config
 source setup.sh
 gmake
 cd ../run
 athena.py AnalysisSkeleton_topOptions_lowpt.py

[shaug@lheppc44] > cat athena.xrsl 

 &(executable=test.sh)
 (inputFiles=("myAnalysis.tar" "myAnalysis.tar"))
 (outputfiles=("/" ""))
 (stdout=test.log)
 (stderr=test.err)
 (jobname=sigves-athena)
 (runTimeEnvironment=APPS/HEP/ATLAS-14.1.0.2)

''' Example without compiling remotely :'''
In order to send the compiled job to the grid we have to include the following directory in the gridfiles.tar file 

 InstallArea/python

as well as the libraries (inside InstallArea/i686-slc4-gcc34-opt/lib/*.so)

 libUserAnalysis.so, 
 libAnalysisUtilsLib.so, 
 libUserAnalysisUtils.so


The resulting tar file will then contain all the custom .so files and the python directory.

In order to make the system use the custom libraries, one needs to set LD_LIBRARY_PATH (for libraries) and PYTHONPATH (for python bindings):

(just after the tar -xvzf command)

 cd gridfiles
 a=$(pwd)
 export LD_LIBRARY_PATH=$a:$LD_LIBRARY_PATH
 export LD_LIBRARY_PATH=$a/python:$LD_LIBRARY_PATH
 export PYTHONPATH=$a/python:$PYTHONPATH


An example bash script which will be executed on the grid: 

 #/bin/bash
 tar -xvf gridfiles.tar
 cd gridfiles
 a=$(pwd)
 export LD_LIBRARY_PATH=$a:$LD_LIBRARY_PATH
 export LD_LIBRARY_PATH=$a/python:$LD_LIBRARY_PATH
 export PYTHONPATH=$a/python:$PYTHONPATH
 athena.py AnalysisSkeleton_topOptions.py