GSI Forum
GSI Helmholtzzentrum für Schwerionenforschung

Home » PANDA » PandaRoot » Bugs, Fixes, Releases » Tutorial for analysis macros on the GRID
Tutorial for analysis macros on the GRID [message #12130] Thu, 23 June 2011 18:07
StefanoSpataro is currently offline  StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino
first-grade participant

From: *to.infn.it
Dear all,
in order to run analysis on the GRID for tracking TDR, I have written some sample files, to use it as tutorial.

First of all, you have to create a collection with the .zip files containing the root file you want to analyze. You can run at present a maximum amount of 60 jobs in parallel, thus a maximum of 60 folders.
In order to create a collection "coll_702_a" with all the sig_output_run702.zip files from runid 702 (pipi), all the sub-folders in 10*:

find -c /panda/user/s/spataro/tdr11/coll/coll_702_a /panda/user/p/pbarprod/tdr11/output/sig/run702/10*/ sig_output_run702.zip


Now we have to create the script doing the job. This should stay in the /panda/user/x/xxx/bin folder (i.e., in my case, /panda/user/s/spataro/bin). I have done a copy of the test one in pbarprod account, that you can use:

cp /panda/user/p/pbarprod/bin/test-ana.sh  /panda/user/s/spataro/bin/test-ana.sh


This script, if you read it, is unzipping the .zip file (containing the root files), renaming the files into the standard evt_XXX_stt.root, and executing the macro run_ana_pipi.C with root. If the output file "finalroot_piplus.root" is not existing then it gives an exit statement (error!).

Toggle Spoiler


The question now is: how to set the input file, the output folder, which script to execute and from where taking the macro. For this, you have to create your jdl, or copy the one I have created for you:

cp /panda/user/p/pbarprod/tdr11/jdl/test-ana.jdl /panda/user/s/spataro/tdr11/test-ana.jdl


Here you set your script, the input data collection, the "input file" (i.e. the macros that will be copied), the output archive (the root file will be archived into one .zip stored in 2 disks, and also the log files will be stored in 1 disk). Split directory means that a subjob will be sent for each directory inside your collection. Finally the output files will be stored into Outputdir. You have to modify it (using "vi" or just "getting" the file in your local pc, updating it and "adding" it again into the grid) because you do not want to retrieve files and store output inside the pbarprod, but in your account, and you want to use your new collection. At the end the jdl could look like:

Toggle Spoiler


And you have only to submit it:

[pgdb2.gla.ac.uk:3307] /panda/user/s/spataro/tdr11/ > submit test-ana.jdl
Jun 23 17:50:11  info	Submitting job '/panda/user/s/spataro/bin/test-ana.sh '...
Jun 23 17:50:11  info	There is no price defined for this job in the jdl. Putting the default '1.0' 
Jun 23 17:50:11  info	*** calling PackMan with arguments list -silent -all
Jun 23 17:50:11  info	Calling directly getListPackages (list -silent -all)
Jun 23 17:50:11  info	Checking the input collection LF:/panda/user/s/spataro/tdr11/coll/coll_702_a
Jun 23 17:50:11  info	Input Box: {run_ana_pipi.C}

    ATTENTION. You just submitted a JDL containing the tag 'OutputArchive'. The OutputFile and OutputArchive
    tags will be dropped in future versions of AliEn. For the moment the old tags work as usual, but
    please update your JDLs in the near future to utilize the 'Output' tag:

    The syntax of the actual entries is still the same, but now you can just mixup files
    and archives, as e.g.:

           Output = { "fileA,fileB,*.abc" , "myArchive:fileC,fileD,*.xyz" } ;


    Thanks a lot!

Jun 23 17:50:15  info	OK, all right!
Jun 23 17:50:15  info	Command submitted (job 1171481)!!
Job ID is 1171481 - 0


You can see how your jobs are going with masterJob command:

[pgdb2.gla.ac.uk:3307] /panda/user/s/spataro/tdr11/ > masterJob 1171481
Jun 23 17:56:21  info	Checking the masterjob of 1171481
Jun 23 17:56:21  info	The job 1171481 is in status: SPLIT
It has the following subjobs:
		Subjobs in DONE: 10 
		Subjobs in STARTED: 1 

In total, there are 11 subjobs


Once all the jobs are DONE (you can check it also here), you have to do a collection with your root files, storing your ntuples, get the collection into your local pc, merging the histogram and show nice results!

I hope this guide is clear enough.

[Updated on: Thu, 23 June 2011 18:07]

Report message to a moderator

Previous Topic: Possible Programming Error(build fails)
Next Topic: PANDAroot simulation with FTOF
Goto Forum:
  


Current Time: Sun Dec 08 12:20:39 CET 2024

Total time taken to generate the page: 0.00686 seconds