Running the tutorial example at BNL Tier3

Table of Contents

• Login to BNL machines
• Location of this analysis example package at BNL
• Env Setup and Package Building
  • Setup of the Release Env
  • Package Building
• Dataset Preparation
• Interactive Job Running at BNL
• Condor Batch Job Running at BNL
• Using Xcache at BNL
  • Using Xcache (gLFN) at BNL

Login to BNL machines

First you need ssh to spar machines at BNL:

$ ssh -Y atlasgw
$ rterm -i

Location of this analysis example package at BNL

You can pull the analysis package under the same github repo directory or copy from the directory at BNL:

/atlasgpfs01/usatlas/data/yesw2000/T3-Example-BNL/
|-- 00-Readme.txt
|-- Condor-Job
|   `-- test-condor.job
`-- src
    |-- CMakeLists.txt
    `-- Exam_JetsPlot.cxx

where you can find: - the text instruction file 00-Readme.txt, from which you can copy/paste the commands used in this tutorial. - The condor job description file test-condor.job - And the source code of this analysis pacakge under subdir src/

Env Setup and Package Building

Setup of the Release Env

First set up the release env by:

setupATLAS
asetup AnalysisBase,21.2.81

Package Building

To build the single package, just run under that dir:

cmake src  # generate Makefile
make       # compile & create executable file

After the package is built, you will see

$ ls -F 00-Readme.txt Makefile CMakeCache.txt bin/ CMakeFiles/ cmake_install.cmake Condor-Job/ src/

and

$ ls bin/
Exam_JetsPlot

To get the usage of the single analysis program, run

bin/Exam_JetsPlot -h

Usage: bin/Exam_JetsPlot  inputRootFile

        A simple example to make a few jet plots

Options:
        -h,--help               Show this help message

Dataset Preparation

To run the analysis, we need an input dataset.

dset=data18_13TeV.00348885.physics_Main.deriv.DAOD_EXOT12.f937_m1972_p3553_tid14278917_00

lsetup rucio
voms-proxy-init -voms atlas
pnfs_ls.py $dset -o dset-inside.clist                       # pls run "pns_ls.py -h" for usage (next section)
pnfs_ls.py $dset --useXRootdOutside -o dset-outside.clist   # for outside access

To get the script pnfs_ls.py usage, just run pnfs_ls.py -h.

Please click the following arrow to see the full usage.

Let us look into the generated file dset-inside.clist $ head -5 dset-inside.clist

# clist file for access within BNL
#site= BNL-OSG2_LOCALGROUPDISK ; dset= data18_13TeV.00348885.physics_Main.deriv.DAOD_EXOT12.f937_m1972_p3553_tid14278917_00
root://dcgftp.usatlas.bnl.gov:1096//pnfs/usatlas.bnl.gov/LOCALGROUPDISK/rucio/data18_13TeV/da/ea/DAOD_EXOT12.14278917._000001.pool.root.1
root://dcgftp.usatlas.bnl.gov:1096//pnfs/usatlas.bnl.gov/LOCALGROUPDISK/rucio/data18_13TeV/32/82/DAOD_EXOT12.14278917._000002.pool.root.1
root://dcgftp.usatlas.bnl.gov:1096//pnfs/usatlas.bnl.gov/LOCALGROUPDISK/rucio/data18_13TeV/86/76/DAOD_EXOT12.14278917._000003.pool.root.1

Interactive Job Running at BNL

Now we can run the job interactively

mkdir Interactive-Job
cd Interactive-Job

Take one input file in "dset-inside.clist"

inputFile=root://dcgftp.usatlas.bnl.gov:1096/pnfs/usatlas.bnl.gov/LOCALGROUPDISK/rucio/data18_13TeV/da/ea/DAOD_EXOT12.14278917._000001.pool.root.1
../bin/Exam_JetsPlot $inputFile > myjob.log 2>&1

It will write out an output file myOutputFile.root You can take a look of the log file myjob.log, containing info of input file and event processing progress.

Let us look into the output root file myOutputFile.root

$ root -l  myOutputFile.root
root [1] .ls
TFile**         myOutputFile.root
 TFile*         myOutputFile.root
  KEY: TH1D     h_njets_raw;1
  KEY: TH1D     h_mjj_raw;1
root [2] h_mjj_raw->Draw();

which will yield the plot

Condor Batch Job Running at BNL

At BNL, the batch queue uses condor system. The working directory of Condor batch jobs is different from current working directory, and you need specify "GetEnv = True" to pass the current env variables.

cd Condor-Job
ls -l $X509_USER_PROXY    # under local disk /tmp/, we need copy/delegate it to the batch machines

Let us take a look of job description file test-condor.job

Please click the following filename to see its content

Run the following command to submit the condor job

$ condor_submit  test-condor.job

Then you can run condor_q to check your job status.

$ condor_q
-- Schedd: spar0103.usatlas.bnl.gov : <130.199.48.19:9618?... @ 08/02/19 13:12:27
 ID       OWNER            SUBMITTED     RUN_TIME ST PRI SIZE CMD
35106.0   yesw2000        8/2  13:12   0+00:00:00 I  0    0.3 Exam_JetsPlot

Total for query: 1 jobs; 0 completed, 0 removed, 1 idle, 0 running, 0 held, 0 suspended 
Total for yesw2000: 1 jobs; 0 completed, 0 removed, 1 idle, 0 running, 0 held, 0 suspended 
Total for all users: 2 jobs; 0 completed, 0 removed, 1 idle, 1 running, 0 held, 0 suspended

After the job is done, we will see

$ ls -l
total 44
-rw-r--r-- 1 yesw2000 usatlas  4367 Aug  2 13:15 myOutputFile.root
-rw-r--r-- 1 yesw2000 usatlas   967 Aug  2 13:15 myjob-running.log
-rw-r--r-- 1 yesw2000 usatlas   335 Aug  2 13:15 myjob.err
-rw-r--r-- 1 yesw2000 usatlas 20934 Aug  2 13:15 myjob.out
-rw-r--r-- 1 yesw2000 usatlas   434 Aug  2 12:18 test-condor.job

There are 4 files created by the batch job: - myjob-running.log, is the condor log file, created right at the job submission and being updated until the job completion. - Other 3 files, myjob.out, myjob.err and myOutputFile.root, created on the batch machine, transferred back after the job completion.

Similarly, you can also look into the output file myOutputFile.root and make the plot.

Using Xcache at BNL

Xcache enables to access data remotely and also to cache them locally for faster access in future.

The Xcache server at BNL is root://xrootd03.usatlas.bnl.gov:1094/.

Let us take the input file used in the SLAC example. At SLAC, the inputFile name for outside access (check the file dset-outside.txt at SLAC) is

inputFile=root://griddev03.slac.stanford.edu:2094//xrootd/atlas/atlaslocalgroupdisk/rucio/data16_13TeV/f9/bd/DAOD_SUSY15.11525262._000003.pool.root.1

For Xcache, we need add the Xcache server prefix with two slash characters, that is,

inputFile=root://xrootd03.usatlas.bnl.gov:1094//root://griddev03.slac.stanford.edu:2094//xrootd/atlas/atlaslocalgroupdisk/rucio/data16_13TeV/f9/bd/DAOD_SUSY15.11525262._000003.pool.root.1
cd T3-Example-BNL/Interactive-Job
../bin/Exam_JetsPlot $inputFile > myjob.log 2>&1

Using Xcache (gLFN) at BNL

Xcache at BNL also supports gLFN (global Logical File Name) access, without the need of knowing the exact path of a given filename.

Let us take the same dataset used in the SLAC example.

Please click the following command (where dset is defined previously) to see the output

Let us to list the filenames in the dataset

$ rucio list-content $dset
+-------------------------------------------------------+--------------+
| SCOPE:NAME                                            | [DID TYPE]   |
|-------------------------------------------------------+--------------|
| data16_13TeV:DAOD_SUSY15.11525262._000003.pool.root.1 | FILE         |
| data16_13TeV:DAOD_SUSY15.11525262._000006.pool.root.1 | FILE         |
...

Let us take the second one file.

inputFile=root://xrootd03.usatlas.bnl.gov:1094//atlas/rucio/data16_13TeV:DAOD_SUSY15.11525262._000003.pool.root.1
../bin/Exam_JetsPlot $inputFile > myjob.log 2>&1

Enclosed is a screenshot of the condor running jobs.

$ condor_q
-- Schedd: spar0103.usatlas.bnl.gov : <130.199.48.19:9618?... @ 08/02/19 13:12:36
 ID       OWNER            SUBMITTED     RUN_TIME ST PRI SIZE CMD
35106.0   yesw2000        8/2  13:12   0+00:00:01 R  0    0.3 Exam_JetsPlot

Total for query: 1 jobs; 0 completed, 0 removed, 0 idle, 1 running, 0 held, 0 suspended 
Total for yesw2000: 1 jobs; 0 completed, 0 removed, 0 idle, 1 running, 0 held, 0 suspended 
Total for all users: 2 jobs; 0 completed, 0 removed, 0 idle, 2 running, 0 held, 0 suspended