run-clara

#!/bin/bash -f

ulimit -u 49152 >& /dev/null
set -e

usage="Usage:  run-clara -y YAML [-h] [-m] [-t #] [-n #] [-o DIR] [-p PREFIX] [-c CLARA_HOME] FILE..."
info='\nRequired Arguments:\n
\tFILE... (input data files)\n
\t-y   YAML file\n
Options:\n
\t-o   output directory (default=.)\n
\t-p   output prefix (default=rec_)\n
\t-c   CLARA installation (default=$CLARA_HOME)\n
\t-t   number of threads (default=2)\n
\t-n   number of events (default=-1)\n
\t-m   merge output files (see dependencies below)\n
\t-l   larger JVM memory requests\n
\t-L   set Java log level (e.g., info, fine, etc.)\n
\t-q   quiet output (overrides -L)\n
\t-v   verbose output (xtrace)\n
\t-h   print this help and exit\n\n
Merging outputs (-m) requires hipo-utils and yq (https://github.com/mikefarah/yq).'

function error() {
    echo -e "\n$usage\n\nERROR:  $@." >&2 && exit 1
}

# Interpret command line:
threads=2
prefix=rec_
java_quiet=false
java_log_level=info
CLARA_USER_DATA=.
while getopts y:o:p:c:t:n:lL:qmvh opt
do
    case $opt in
        y) yaml=$OPTARG ;;
        o) CLARA_USER_DATA=$OPTARG ;;
        p) prefix=$OPTARG ;;
        c) CLARA_HOME=$OPTARG ;;
        t) threads=$OPTARG && echo $threads | grep -q -E '^[0-9]+$' || error "-t must be an integer, threads" ;;
        n) nevents="-e $OPTARG" && echo "$nevents" | grep -q -E '^-e [0-9]+$' || error "-n must be an integer, events" ;;
        l) large=1 ;;
        L) java_log_level=$OPTARG ;;
        m) merge=1 ;;
        q) java_quiet=true ;;
        v) set -o xtrace ;;
        h) echo -e "\n$usage" && echo -e $info && exit 0 ;;
    esac
done
shift $((OPTIND-1))
inputs=$@

# Configure JVM memory settings (overridedable via $JAVA_OPTS):
gb_init=$((threads))
gb_max=$((threads+2))
[ $threads -gt 1 ] && ! [ -z ${large+x} ] && gb_max=$((threads*2))
java_opts="-Xms${gb_init}g -Xmx${gb_max}g"

# Check configuration:
[ $# -lt 1 ] && error "Input data files are required"
[ -z ${yaml+x} ] && error "-y YAML is required"
[ -f $yaml ] && [ -r $yaml ] || error "YAML file does not exist:  $yaml"
[ -z ${CLARA_HOME+x} ] && error "-c must be specified or \$CLARA_HOME set"
[ -d $CLARA_HOME ] || error "Invalid CLARA_HOME:  $CLARA_HOME"
[ $threads -eq 0 ] && threads=`grep -c ^processor /proc/cpuinfo`
! [ -z ${merge+x} ] && ! command -v hipo-utils >& /dev/null && error "Merging requested, but hipo-utils is not in \$PATH"
yaml=$(cd $(dirname $yaml) && pwd)/$(basename $yaml)

# Create the environment variables and directories required by CLARA:
[ -e $CLARA_USER_DATA ] && echo "WARNING:  Using existing directory:  $CLARA_USER_DATA" >&2
mkdir -p $CLARA_USER_DATA || error "Cannot create -o output directory: $CLARA_USER_DATA"
mkdir -p $CLARA_USER_DATA/log $CLARA_USER_DATA/config $CLARA_USER_DATA/data/output
export CLARA_USER_DATA=$(cd $CLARA_USER_DATA && pwd)
export CLARA_HOME=$(cd $CLARA_HOME && pwd)
export CLAS12DIR=$CLARA_HOME/plugins/clas12
unset CLARA_MONITOR_FE

# Generate the file for CLARA containing a list of file basenames:
rm -f $CLARA_USER_DATA/filelist.txt && touch $CLARA_USER_DATA/filelist.txt
for x in $inputs
do
    test -f $x && test -r $x || error "Invalid input file:  $x"
    echo $(basename $x) >> $CLARA_USER_DATA/filelist.txt
    test -f $CLARA_USER_DATA/$(basename $x) || ln -sf $(cd $(dirname $x) && pwd)/$(basename $x) $CLARA_USER_DATA
done
[ $(cat $CLARA_USER_DATA/filelist.txt | wc -l) -gt 0 ] || error "Found no input files"

# Set some JVM options:
JAVA_OPTS="$java_opts $JAVA_OPTS -XX:+IgnoreUnrecognizedVMOptions"
JAVA_OPTS="$JAVA_OPTS -Djava.io.tmpdir=$CLARA_USER_DATA -Dorg.sqlite.tmpdir=$CLARA_USER_DATA"

# Set logger and verbosity
JAVA_OPTS="$JAVA_OPTS -Djava.util.logging.manager=org.jlab.logging.SplitLogManager"
if $java_quiet; then
  JAVA_OPTS="$JAVA_OPTS -Djava.util.logging.config.file=$CLAS12DIR/etc/logging/logging.properties"
else
  java_log_level_upcase=$(echo $java_log_level | tr '[:lower:]' '[:upper:]')
  case ${java_log_level_upcase} in
    SEVERE|WARNING|INFO|CONFIG|FINE|FINER|FINEST|ALL|OFF)
      JAVA_OPTS="$JAVA_OPTS -D.level=${java_log_level_upcase}"
      ;;
    *)
      error "unknown Java log level '$java_log_level'"
      ;;
  esac
fi
export JAVA_OPTS

function get_host_ip() {
    if command -v ip >/dev/null 2>&1
    then
        ip route get 1 | awk '{print $7; exit}' && return 0
    elif command -v ifconfig >/dev/null 2>&1
    then
        while IFS=$': \t' read -r -a line
        do
            if [ -z "${line%inet}" ]
            then
                ip=${line[${#line[1]}>4?1:2]}
                [ "${ip#127.0.0.1}" ]
                echo $ip && return 0
            fi
        done< <(LANG=C ifconfig)
    fi
    return 1
}
function get_dpe_port() {
    local ports
    ports=$(seq 7000 20 8000)
    command -v shuf >/dev/null 2>&1 && ports=$(echo "$ports" | shuf)
    for port in $ports
    do
        local ctrl_port=$((port + 2))
        if ! eval "exec 6<>/dev/tcp/127.0.0.1/$ctrl_port" 2> /dev/null
        then
            echo $port
            return 0
        fi
    done
    return 1
}

# Finally, run CLARA:
if [ $(uname) == "Darwin" ]
then
    ip=$(get_host_ip) || error "Unknown IP address"
    port=$(get_dpe_port) || error "Unknown DPE port"
    set -v
    $CLARA_HOME/bin/j_dpe \
        --host $ip --port $port \
        --session recon --max-cores $threads \
        --max-sockets 5120 --report 5 \
        2>&1 | tee $CLARA_USER_DATA/log/dpe.log &
    set +v
    #echo "Sleeping 7 ......." && sleep 7
    unset JAVA_OPTS
    set -v
    $CLARA_HOME/bin/clara-orchestrator \
        -F -f ${ip}%${port}_java -s recon \
            -i $CLARA_USER_DATA -o $CLARA_USER_DATA -z $prefix \
            -p $threads -t $threads $nevents \
            $yaml $CLARA_USER_DATA/filelist.txt
    set +v
else
    set -v
    $CLARA_HOME/lib/clara/run-clara \
    -i $CLARA_USER_DATA \
    -o $CLARA_USER_DATA \
    -z $prefix \
    -x $CLARA_USER_DATA/log \
    -t $threads \
    $nevents \
    -s recon \
    $yaml $CLARA_USER_DATA/filelist.txt
    set +v
fi

# Merge outputs:
if ! [ -z ${merge+x} ]
then
    if grep -q org.jlab.jnp.grapes $yaml >& /dev/null
    then
        for id in $(yq .configuration.services.*.id $yaml | sort -n | uniq)
        do
            hipo-utils -merge -o $CLARA_USER_DATA/$prefix$id.hipo $CLARA_USER_DATA/$prefix*$id.hipo
        done
    else
        outfiles=$(sed "s#^#$CLARA_USER_DATA/$prefix#" $CLARA_USER_DATA/filelist.txt)
        hipo-utils -merge -o $CLARA_USER_DATA/$prefix.hipo $outfiles
    fi
fi