Using the ATLAS Software at Liverpool

Getting Started

• Log on to one of the interactive nodes (note: SLC5 is required for all recent ATHENA releases >=15.6.3)
  • SLC5 64 bit: iota, theta, ... (or use hepshell64)
  • SLC4 32 bit: alpha, beta, ... (or use hepshell)

• The ATLAS software is installed in /batchsoft/atlas/, especially you may need /batchsoft/atlas/athena/<version.number>

Calculate Luminosity Locally

The following lines are needed in order to calculate the luminosity using the iLumiCalc.exe locally:

 export CORAL_AUTH_PATH=/batchsoft/atlas/athena/<version>/AtlasCore/<version>/InstallArea/XML/AtlasAuthentication
 export CORAL_DBLOOKUP_PATH=$CORAL_AUTH_PATH

where version should be replaced by the athena version that you're running in (e.g. 15.6.9).

the required releases for the luminosity calculations can be found from https://twiki.cern.ch/twiki/bin/viewauth/Atlas/GoodRunsLists

New simplified ATHENA setup

Using the AtlasSetup package the user doesn't have to bother with all the cmt nastiness. This pacakge is replacing AtlasLogin centrally from release ~16.2.0 on. Long documentation at:

https://twiki.cern.ch/twiki/bin/view/Atlas/AtlasSetup

The old cmt-based setup will continue to work as it has done before for releases <16.2.0, so you don't need to change if you don't want to. The advantage of this is that we can setup a site-wide configuration and very little extra user configuration.

The simplest is to source the following setup script automatically in your .bashrc, i.e. add:

# atlas bashrc
[[ -f /batchsoft/atlas/atlas_bashrc ]] && . /batchsoft/atlas/atlas_bashrc

Alternatively, add the following lines to your .bashrc/.zshenv:

# Setup AtlasSetup and export site-wide config
export AtlasSetup=/batchsoft/atlas/athena/AtlasSetup
alias asetup='source $AtlasSetup/scripts/asetup.sh'
export AtlasSetupSite=/batchsoft/atlas/athena/asetup.site

To setup a release one then just has to do e.g.

asetup 15.6.9.8

Note that you don't have to specify AtlasProduction for caches as it knows about this already. By default it will assume that your TestArea is wherever you source the asetup script from. If you want to specify a testarea, or any other configuration for that matter, you need to create a file in your home directory called .asetup containing something like the following:

[defaults]
testarea = /user1/gwilliam/AtlasWork15

These setup files are based on INI files (rather than cmt) and, in general, this is the only line you need to specify. One other change to note is that this is setup by default to use the same TestArea for all caches of a particular release, e.g. it will use 15.6.9 for any 15.6.9.Y, which saves moving TestAreas too often. If you want to override this you can add to your .asetup the following line under [defaults]:

testtype="<release>"
 

NB DQ2 does not work after athena has been set up. To use DQ2, type the following in a clean shell

grid_setup

Running Interactively

• Log-on to an interactive node and do

  source ~/cmthome/setup.sh -tag=AtlasProduction,< version >

• Get the "Hello World" options

  get_files HelloWorldOptions.py

• Run athena:

  athena HelloWorldOptions.py

Running in Batch

• Make sure you are logged onto the batch front end hepcluster
• Write a script, AthenaTest.sh, to setup the athena environment and run your job:

  echo "Sourcing Athena Setup"
  source ~/cmthome/setup.sh -tag=AtlasProduction,15.6.0.1
  echo "Running athena"
  athena.py ~/HelloWorldOptions.py
   

• If the directory /hepstore/store?/$USER does not already exist, create this directory (where store1..store5 are available, but all use the same volume)

  mkdir /hepstore/store?//$USER

• Submit using qsub

  qsub -o /batchstore/$USER -e /batchstore/$USER -m ea -N JobName AthenaTest.sh
   

• For a full list of options see the man page for qsub. The ones used here are:

  -o directory to save stdout to (don't use your user area)
  -e directory to save stderr to
  -m send a mail when the job begins (b), ends (e), or aborts (a)
  -N set job name
   

• Check the status of your jobs using

   qstat -u $USER 
  

Running in Batch using Ganga

• You can run in batch using ganga. To do this you need to edit your .gangarc by adding the section

  [PBS]

   submit_str = cd %s; qsub %s %s %s %s submit_res_pattern = ^(?P<id>\d*).hepcluster.ph.liv.ac.uk 

  and ensure the atlas software directory is correctly set

   [Athena] ATLAS_SOFTWARE = /batchsoft/atlas/athena 

  and change your gangadir to be on /scratch/<username>. E.g.

  gangadir =/scratch/jkretz/gangadir/ 

  This is necessary because the stdout is prevented from going to your user directory

• You can then run the athena hello world job in ganga via

  j=Job() j.application=Athena() j.application.atlas_release='15.6.0' j.application.option_file='/user2/jkretz/testarea/15.6.0.1/PhysicsAnalysis/AnalysisCommon/UserAnalysis/run/HelloWorldOptions.py' j.backend=PBS() j.backend.queue = 'medium' j.submit() 

Using Atlantis

• On any of the interactive nodes you can run atlantis from the athena kit:

  java -jar /batchsoft/atlas/athena/15.6.0/AtlasAnalysis/15.6.0/InstallArea/share/AtlantisJava/atlantis.jar 

Using Liverpool CVS

• LiverpoolCVS

-- CarlGwilliam - 19 Feb 2007