- setup on lxplus
your account should be able to run athena "hello world" as discribed in the workbook ["https://twiki.cern.ch/twiki/bin/view/Atlas/WorkBookRunAthenaHelloWorld"]
- login to lxplus
- start athena
> source ~/cmthome/setup.sh -tag=12.0.4
- start the grid
> source /afs/cern.ch/project/gd/LCG-share/sl3/etc/profile.d/grid_env.sh
> source /afs/cern.ch/project/gd/LCG-share/sl3/etc/profile.d/grid_env.sh
> export LFC_HOST=lfc-atlas.cern.ch
> export LCG_CATALOG_TYPE=lfc
> voms-proxy-init
- create a long term proxy (recommended for mc production, default 1 week)
> myproxy-init
- check with
> myproxy-info
- the long term proxy lies on "myproxy.cern.ch" and is automatically used by ganga
- start dq2 (dataset mangagement)
> source /afs/usatlas.bnl.gov/Grid/Don-Quijote/dq2_user_client/setup.sh.CERN
- start ganga (4.2.7) as atlas user in graphical user interface with
> /afs/cern.ch/sw/ganga/install/slc3_gcc323/4.2.7/bin/ganga --config-path=GangaAtlas/Atlas.ini --gui
> /afs/cern.ch/sw/ganga/install/slc3_gcc323/4.2.7/bin/ganga --config-path=GangaAtlas/Atlas.ini --gui
- type "y" to create a default ganga config file ".gangarc"
- ganga should start
- open a second shell, and type
> source ~/cmthome/setup.sh -tag=12.0.4
> source /afs/cern.ch/project/gd/LCG-share/sl3/etc/profile.d/grid_env.sh
> source /afs/cern.ch/project/gd/LCG-share/sl3/etc/profile.d/grid_env.sh
> export LFC_HOST=lfc-atlas.cern.ch
> export LCG_CATALOG_TYPE=lfc
> voms-proxy-init
> source /afs/usatlas.bnl.gov/Grid/Don-Quijote/dq2_user_client/setup.sh.CERN
> source /afs/usatlas.bnl.gov/Grid/Don-Quijote/dq2_user_client/setup.sh.CERN
- change into a run directotory
> cd testarea/12.0.4/PhysicsAnalysis/AnalysisCommon/UserAnalysis/run
setup on atlhltX (X=2,3,4)
- first you have to install ganga
- login into atlhltX and change to /usr1/scratch/userscratch/USERNAME
- you can download the ganga install script here
- or copy into your disk with
> cp /afs/cern.ch/user/v/volkmann/public/ganga-install .
- install 4.2.7 with options
> python ganga-install --prefix=/usr1/scratch/userscratch/USERNAME/Ganga/ --extern=GangaAtlas,GangaGUI,GangaPlotter
- the install script shows the file to start ganga
- your account should be able to run athena "hello world"
- start athena
> source /usr1/scratch/userscratch/USERNAME/cmthome/setup.sh -tag=12.0.3
> source /usr1/scratch/userscratch/USERNAME/cmthome/setup.sh -tag=12.0.3
- start the grid
> source /afs/desy.de/group/grid/UI/GLITE-3_0_4/etc/profile.d/grid_env.sh
> source /afs/desy.de/group/grid/UI/GLITE-3_0_4/etc/profile.d/grid_env.sh
> export LFC_HOST=lfc-atlas.cern.ch
> export LCG_CATALOG_TYPE=lfc
> voms-proxy-init
- log with another shell into lxplus and create a long term proxy with "myproxy-init" as described above (myproxy-init at atlhltX fails)
- start dq2
> source /afs/ifh.de/project/grid/dq2/setup.sh
uncomment the line "MyProxyServer = myproxy.cern.ch" in the ganga config file "~/.gangarc"
- start ganga for atlas with graphical user interface
> /usr1/scratch/userscratch/USERNAME/Ganga/install/slc3_gcc323/4.2.7/bin/ganga --config-path=GangaAtlas/Atlas.ini --gui
- open anther shell in atlhltX, start athena, grid, dq2, change into run directory (as discribed before)
hello world with Ganga in graphical user interface
you should contain "HelloWorldOptions.py" in your run directory
- click on "new" in the ganga window to create a new job
- click on application and change from "Executable" to "Athena" in the drop down box on the right side
- expand the application menu
click on "option file" and browse on your disk to "HelloWorldOptions.py" to load it into ganga
- click on "max events" and write "10"
- click on "backend" and change from "local" to "LCG"
- click on "submit" button
- inspect the status of the job from "submitted", "running" to "completed", hopefully not to "failed"
- retrieve the output by clicking on the job with the right button and choose "retrieve output", click on the file you want to read
- you can also inspect input/output directly in ~/gangadir/workspace/...
generating events (1. step in the chain)
- the normal way would be to download in the second shell the atlas production transform archive, actual releases can be found here
- by default you have to create minimum 5000 events, but for educational test and to save time, we create 3 events,
- creating more than 3 is simple and strait forward (but takes longer time)
- you can change the default of 5000 events as discribed here
- to save time download a file already changed down to 3 events (default is only changed for 1 example: "DC3.005711.PythiaB_bbmu6mu4X.py")
> cp "/afs/cern.ch/user/v/volkmann/public/AtlasProduction_12_0_31_8_noarch.tar.gz" .
- click on new in the ganga window
- change "application" to "AthenaMC"
- type into atlas release "12.0.31"
- type into eventgen_job_option "DC3.005711.PythiaB_bbmu6mu4X.py"
- type into firstevent "1"
- make sure that mode is "evgen"
- type into number_events_job "3"
- type into process_name something meaningful like "bbmu6mu4X"
- type into production_name another meaningful thing like "bphys"
- change the random seed to a 9 digit number
- type in a 6-digit run number like "000001"
browse the transform script "AtlasProduction_12_0_31_8_noarch.tar.gz" into ganga
- type into transform_script "csc_evgen_trf.py"
- click on backend and choose "LCG"
- expand LCG and choose one of the following computing elements
** tbn20.nikhef.nl:2119/jobmanager-pbs-qlong
** lcg-ce0.ifh.de:2119/jobmanager-lcgpbs-atlas
** grid-ce1.desy.de:2119/jobmanager-lcgpbs-atlas
- (you find all by typing "lcg-infosites --vo atlas ce" into the shell)
- click on outputdata and select "DQOutputDataset"
- click on splitter and choose "AthenaMCSplitterJob"
- click on numsubjobs and type "3"
- click on nsubjobs_inputfile and type also "3"
- submit the job
- inspect the output in the other open shell with dq2 if the 3 jobs reach "completed" (sometimes dq2 doesn't work at atlhltX)
- you find the output datasetname in ganga
> dq2_ls -g datasetname
- you should see the three root files (each has 1 event in this example)
simulating and digitizing events (2. and 3. step in the chain)
- create a new job, and choose again "AthenaMC" in the application form
- type into the form the atlas_release, prozess_name, production_name and run_number of the previous generation job
- type into first_event "1"
- type into number_events_job "1"
- choose a 9-digit random seed
browse the job transform archieve again into the job "AtlasProduction_12_0_31_8_noarch.tar.gz"
- type into transform_skript "csc_simul_trf.py"
- choose in backend "LCG" and type a ce into it (as previously mentioned)
- click on inputdata and choose "DQ2OutputDataset" and expand the menu
- type into datasetname the name of the dataset from the previous job
- click on outputdata and choose "DQ2OutputDataset"
- choose in Splitter "AthenaMCSplitterJob" and expand
- type into n_subjobs_inputfile "1"
- type into number_input_jobs "3"
- type into numsubjobs "3"
- submit the job and inspect the result with dq2_ls -g datasetname
reconstructing events (4. step in the chain)
- everything remais the same to simulation exept:
- change mode to "recon"
- type into transform_skript "csc_reco_trf.py"
- type into the inputdata the corresponding name of your simulation dataset
- check the output dataset with "dq2_ls -g datasetname"
- download the dataset with "dq2_get datasetname" and inspect with root
links