justIN           Dashboard       Workflows       Jobs       AWT       Sites       Storages       Docs       Login

Workflow 11250, Stage 1

Priority50
Processors4
Wall seconds80000
Image/cvmfs/singularity.opensciencegrid.org/fermilab/fnal-wn-sl7:latest
RSS bytes6815744000 (6500 MiB)
Max distance for inputs100.0
Enabled input RSEs CERN_PDUNE_EOS, DUNE_CA_SFU, DUNE_CERN_EOS, DUNE_ES_PIC, DUNE_FR_CCIN2P3_DISK, DUNE_IN_TIFR, DUNE_IT_INFN_CNAF, DUNE_UK_GLASGOW, DUNE_UK_LANCASTER_CEPH, DUNE_UK_MANCHESTER_CEPH, DUNE_US_BNL_SDCC, DUNE_US_FNAL_DISK_STAGE, FNAL_DCACHE, FNAL_DCACHE_STAGING, FNAL_DCACHE_TEST, MONTECARLO, NIKHEF, PRAGUE, QMUL, RAL-PP, RAL_ECHO, SURFSARA, T3_US_NERSC
Enabled output RSEs CERN_PDUNE_EOS, DUNE_CA_SFU, DUNE_CERN_EOS, DUNE_ES_PIC, DUNE_FR_CCIN2P3_DISK, DUNE_IN_TIFR, DUNE_IT_INFN_CNAF, DUNE_UK_GLASGOW, DUNE_UK_LANCASTER_CEPH, DUNE_UK_MANCHESTER_CEPH, DUNE_US_BNL_SDCC, DUNE_US_FNAL_DISK_STAGE, FNAL_DCACHE, FNAL_DCACHE_STAGING, FNAL_DCACHE_TEST, NIKHEF, PRAGUE, QMUL, RAL-PP, RAL_ECHO, SURFSARA, T3_US_NERSC
Enabled sites BR_CBPF, CA_SFU, CA_Victoria, CERN, CH_UNIBE-LHEP, CZ_FZU, ES_CIEMAT, ES_PIC, FR_CCIN2P3, IT_CNAF, NL_NIKHEF, NL_SURFsara, UK_Bristol, UK_Brunel, UK_Durham, UK_Edinburgh, UK_Glasgow, UK_Imperial, UK_Lancaster, UK_Liverpool, UK_Manchester, UK_Oxford, UK_QMUL, UK_RAL-PPD, UK_RAL-Tier1, UK_Sheffield, US_BNL, US_Colorado, US_FNAL-FermiGrid, US_FNAL-T1, US_Michigan, US_NotreDame, US_PuertoRico, US_SU-ITS, US_Swan, US_UChicago, US_UConn-HPC, US_UCSD, US_Wisconsin
Scopeusertests
Events for this stage

Output patterns

 DestinationPatternLifetimeFor next stageRSE expression
1Rucio usertests:pdvd-cosmics-test-fnal-w11250s1p1prod_*_reco.*7776000False

Environment variables

NameValue
CAMPAIGNpdvd-cosmics-test
DUNESW_TAR/cvmfs/fifeuser1.opensciencegrid.org/sw/dune/7e3cda1e39fdbda22381cdc6cee32b4981bcf6b6
JSONFILEpdvd_cosmics_base_meta.json
MCJob1
NEVENTS10
pipyaml1
UTIL_TAR/cvmfs/fifeuser1.opensciencegrid.org/sw/dune/c1e48d5b35887390d054ba5c1d878fbedb09ee9a
YAMLFILEpdvd_cosmics_gen_nopds.yaml

File states

Total filesFindingUnallocatedAllocatedOutputtingProcessedNot foundFailed
10000047594800

Job states

TotalSubmittedStartedProcessingOutputtingFinishedNotusedAbortedStalledJobscript errorOutputting failedNone processed
151200475140303718020
Files processed00100100200200300300400400Dec-18 14:00Dec-18 15:00Dec-18 16:00Dec-18 17:00Dec-18 18:00Dec-18 19:00Dec-18 20:00Dec-18 21:00Dec-18 22:00Files processedBin start timesNumber per binCA_VictoriaCA_SFUUK_QMULUK_OxfordNL_NIKHEFUK_ImperialUS_FNAL-FermiG…US_FNAL-FermiGridIT_CNAFUK_SheffieldUS_UChicagoES_PICCZ_FZUCERNUS_UCSDUK_ManchesterUK_BrunelUS_FNAL-T1UK_RAL-PPDFR_CCIN2P3

RSEs used

NameInputsOutputs
MONTECARLO10180
DUNE_US_BNL_SDCC01593
DUNE_US_FNAL_DISK_STAGE0531
DUNE_UK_MANCHESTER_CEPH0412
DUNE_CERN_EOS0308
QMUL0252
NIKHEF0248
DUNE_IT_INFN_CNAF0113
DUNE_CA_SFU099
DUNE_ES_PIC080
PRAGUE060
RAL-PP028
DUNE_FR_CCIN2P3_DISK04
DUNE_UK_GLASGOW04

Stats of processed input files as CSV or JSON, and of uploaded output files as CSV or JSON (up to 10000 files included)

File reset events, by site

SiteAllocatedOutputting
ES_PIC30
CA_SFU310
NL_NIKHEF10
CA_Victoria01

Jobscript

#!/bin/bash
#

#These must be defined
#if [ -z $FCL_FILE ]; then
#  echo "Fatal Must provide FCL_FILE env var"
#  exit 1
#fi

if [ -z $UTIL_TAR ]; then
  echo "Fatal must provide UTIL_TAR env var"
  exit 1
fi
export PYTHONPATH=$UTIL_TAR:$PYTHONPATH

source /cvmfs/dune.opensciencegrid.org/products/dune/setup_dune.sh
setup metacat
export METACAT_SERVER_URL=https://metacat.fnal.gov:9443/dune_meta_prod/app
export METACAT_AUTH_SERVER_URL=https://metacat.fnal.gov:8143/auth/dune

if [ -z ${JUSTIN_PROCESSORS} ]; then
  JUSTIN_PROCESSORS=1
fi

echo "Justin processors: ${JUSTIN_PROCESSORS}"

export TF_NUM_THREADS=${JUSTIN_PROCESSORS}   
export OPENBLAS_NUM_THREADS=${JUSTIN_PROCESSORS} 
export JULIA_NUM_THREADS=${JUSTIN_PROCESSORS} 
export MKL_NUM_THREADS=${JUSTIN_PROCESSORS} 
export NUMEXPR_NUM_THREADS=${JUSTIN_PROCESSORS} 
export OMP_NUM_THREADS=${JUSTIN_PROCESSORS}  

echo "Justin specific env vars"
env | grep JUSTIN
now=$(date -u +"%Y%m%dT%H%M%SZ")
jobid=`echo "${JUSTIN_JOBSUB_ID:-1}" | cut -f1 -d'@' | sed -e "s/\./_/"`
stageid=${JUSTIN_STAGE_ID:-1}
a=1
 if [ -n "$DUNESW_TAR" ]; then
   stat ${DUNESW_TAR}
   if [ $? -ne 0 ]; then
     echo "Failed to stat $DUNESW_TAR. Exiting safely"
     exit 0
   fi
 
   export PRODUCTS=$DUNESW_TAR:$PRODUCTS 
   echo "Set dunesw to $DUNESW_TAR"
 fi

 if [ -n "$WIRECELL_TAR" ]; then
   stat ${WIRECELL_TAR}
   if [ $? -ne 0 ]; then
     echo "Failed to stat $WIRECELL_TAR. Exiting safely"
     exit 0
   fi
 
   export PRODUCTS=$WIRECELL_TAR:$PRODUCTS 
   echo "Set wirecell to $WIRECELL_TAR"
 fi

 #Setup recent lar software suite
 DUNESW_VERSION=${DUNESW_VERSION:-v10_13_00d00}
 setup dunesw \
    "${DUNESW_VERSION}" \
    -q "${DUNE_QUALIFIER:-e26:prof}"
 
 setup_exit=$?
 if [ $? -ne 0 ]; then
   echo "Failed to setup dunesw $DUNESW_VERSION $DUNE_QUALIFIER"
   exit $setup_exit
 fi
 
 echo "DUNESW loc:"
 ups active | grep dunesw

 echo "WIRECELL loc:"
 ups active | grep wirecell
 
 export FHICL_FILE_PATH=${DUNESW_TAR}:${FHICL_FILE_PATH}
 export FW_SEARCH_PATH=${DUNESW_TAR}:${FW_SEARCH_PATH}
 
 echo "printing env"
 env
 echo "DUNE specific env vars"
 env | grep DUNE

# User can specify to pip install pyyaml
if [ -n "$pipyaml" ]; then
  echo "Installing yaml"
  pip install --user pyyaml
  echo "Done"
# elif [ -n "${YAML_TAR}" ]; then
#   echo "Using venv at ${YAML_TAR}"
#   source ${YAML_TAR}/yaml_venv/bin/activate
fi


 echo "Will use justin-get-file"
 DID_PFN_RSE=`$JUSTIN_PATH/justin-get-file`
 if [ "${DID_PFN_RSE}" == "" ] ; then
   echo "Could not get file"
   exit 0
 fi
 pfn=`echo ${DID_PFN_RSE} | cut -f2 -d' '`
 did=`echo ${DID_PFN_RSE} | cut -f1 -d' '`
 
 input_filename=`echo $did | cut -f2 -d':'`
 echo "input file: $input_filename"

 echo "jobsub_id: ${JUSTIN_JOBSUB_ID:-1}"
 
 #Get run/subrun from parent if using input -- else from justin job ID/process"
 if [ -z "$MCJob" ]; then
   echo "Getting run subrun from $did"
   run_subrun=`python -m beam_job_utils get_run_subrun -i $did`
   get_run_ret=$?
   if [ $get_run_ret -ne 0 ]; then
     echo "error in get_run_subrun"
   fi
 
   echo $run_subrun
   run=`echo $run_subrun | cut -f1 -d' '`
   subrun=`echo $run_subrun | cut -f2 -d' '`
 else
   run=`echo $jobid | cut -f1 -d'_'`  ##TODO -- check order here
   subrun=`echo $jobid | cut -f2 -d'_'`
 fi
 nevents=${NEVENTS:--1} #Set from env override or -1 
 NSKIP=${NSKIP:-0} 
 echo "nevents: $nevents"
 
 overrides="core.data_tier=full-reconstructed \
  core.application.version=${DUNESW_VERSION} \
  core.application.name=reco \
  core.application.family=dunesw \
  core.application=dunesw.reco \
 "

#User specifies if they'd like to use an input file or not 
if [ -n "$MCJob" ]; then
  inputflag=""
else
  inputflag="-i $pfn --parent $did"
fi

echo "input flag: $inputflag"

starttime=`date +"%s"`.0

# TODO -- metadata file configurable
python -m beam_job_utils run_job \
      --json ${UTIL_TAR}/${JSONFILE} \
      --overrides ${overrides} \
      --run $run \
      --subrun $subrun \
      --yaml ${UTIL_TAR}/${YAMLFILE} \
      --nevents $nevents \
      $inputflag

endtime=`date +"%s"`.0

#output_g4_files=`ls *_g4_stage2.root`
output_reco_files=`ls *_reco.root`

#echo "Forming g4 metadata"
#for FILE in ${output_g4_files[@]}
#do
#        echo "  G4 file: $FILE"

        #Retrieve the key in output filename that corresponds to YAML config. stage
 #       IFS=_ read -ra substrs <<< $(echo ${FILE} | cut -d '.' -f 1)
#        key_found='false'
#        KEY='g4_stage2'
#        iter=0
#        declare -i nsub=${#substrs[@]}-1
#        for substr in ${substrs[@]}; do
#                if [ $key_found == 'true' ]; then
#                        KEY+=$substr
#                        if [ $iter -lt ${nsub} ]; then
#                                KEY+='_'
#                        fi      
#                fi
#                if [ $substr == 'stage2' ]; then
#                        key_found='true'
#                fi      
#                let iter=$iter+1
#        done    
#        echo "    KEY: ${KEY}"
#
#        #Retrieve the FHICL file that corresponds to the output file key
#        FCL=$(grep ${UTIL_TAR}/${YAMLFILE} -e "${KEY}:" -A 1 | grep -e "fcl" | cut -d ':' -f 2)
#        echo "    FCL: ${FCL}"
#
#        python -m meta_maker --start_time $starttime --end_time $endtime --file_format "artroot" \
#                     --app_family "dunesw" --app_name "reco" --app_version ${DUNESW_VERSION} \
#                     --data_tier "detector-simulated" --get_events \
#                     --campaign $CAMPAIGN \
#                     --runs 1 --subruns ${subrun} --fcl $FCL \
#                     -f "${JUSTIN_SCOPE}:${FILE}" -j "${FILE}.json"
#        if [ $? -ne 0 ]; then
#          echo "  Error in g4 metadata"
#          exit 1
#        fi
#        echo "  Ran successfully"
#        cat ${FILE}.json
#done

echo "Forming reco metadata"
for FILE in ${output_reco_files[@]}
do
        echo "  subrun: ${subrun} -> reco file: $FILE"

        #Retrieve the key in output filename that corresponds to YAML config. stage
        IFS=_ read -ra substrs <<< $(echo ${FILE} | cut -d '.' -f 1)
        key_found='false'
        KEY='reco'
        iter=0
        declare -i nsub=${#substrs[@]}-1
        for substr in ${substrs[@]}; do
                if [ $key_found == 'true' ]; then
                        KEY+=$substr
                        if [ $iter -lt ${nsub} ]; then
                                KEY+='_'
                        fi
                fi
                if [ $substr == 'reco' ]; then
                        key_found='true'
                fi
                let iter=$iter+1
        done
        echo "    KEY: ${KEY}"

        #Retrieve the FHICL file that corresponds to the output file key
        FCL=$(grep ${UTIL_TAR}/${YAMLFILE} -e "${KEY}:" -A 1 | grep -e "fcl" | cut -d ':' -f 2)
        echo "    FCL: ${FCL}"
	
	python -m meta_maker --start_time $starttime --end_time $endtime --file_format "artroot" \
                     --app_family "dunesw" --app_name "reco" --app_version ${DUNESW_VERSION} \
                     --data_tier "full-reconstructed" --get_events \
                     --campaign $CAMPAIGN --runs [1] --subruns [${subrun}] \
                     --fcl $FCL \
                     -f "${JUSTIN_SCOPE}:${FILE}" -j "${FILE}.json"

        if [ $? -ne 0 ]; then
          echo "  Error in reco metadata"
          exit 1
        fi
        echo "  Ran successfully"
        cat ${FILE}.json
        let subrun=${subrun}+1
done



exitcode=$?
if [ $exitcode -ne 0 ]; then
  echo "Error running. Exiting with ${exitcode}"
  exit $exitcode
fi

echo "$pfn" > justin-processed-pfns.txt
justIN time: 2025-12-18 21:08:01 UTC       justIN version: 01.05.03