Assumes SSH access allowed between nodes of cluster

Run Distributed TF

#!/bin/bash
##Author: Sahil Badyal <[email protected]>
usage()
{
        echo "usage: sysinfo_page [[[-f file ] [-i]] | [-h]]"
}

if [ "1" == "" ]; then
    usage
    exit
fi
##Main
## bash run_distributed.sh -o <o path> -i <i path> -bs 512 -s 800 -e 4000  -f 477
while [ "$1" != "" ]; do
    case $1 in
        -o | --output )         shift
                                out_dir=$1
                                ;;
        -i | --input )          shift
                                input_dir=$1
                                ;;
        -f | --feature_lene)    shift
                                flen=$1
                                ;;
        -bs | --batch-size )    shift
                                batch_size=$1
                                ;;
        -s | --save-every )     shift
                                save=$1
                                ;;
        -e | --eval-every )     shift
                                eval_=$1
                                ;;
        -pb | --prefetch-buffer ) shift
                                pb=$1
                                ;;
        -sb | --shuffle-buffer ) shift
                                sb=$1
                                ;;
        -h | --help )           usage
                                exit
                                ;;
        * )                     usage
                                exit 1
    esac
    shift
done
chief=$(echo $TF_CONFIG | jq '.cluster' | jq '.chief'  | jq '.[]'  | sed 's/:2222//g' | tr -d '"')
ips=$(echo $TF_CONFIG | jq '.cluster' | jq '.worker' | jq '.[]'  | sed 's/:2222//g' | tr -d '"')
tot_workers=$(echo $TF_CONFIG | jq '.cluster' | jq '.worker' | jq '.[]'  | wc -l)
tot_workers=$((tot_workers+1))
ips_ps=$(echo $TF_CONFIG | jq '.cluster' | jq '.ps'   | jq '.[]'  | sed 's/:2222//g' | tr -d '"')
read -a host <<< $ips
read -a ps <<< $ips_ps
k=0
python <path to file> -rdb -bs $batch_size -i $input_dir -pb $pb -sb $sb -o $out_dir -fl $flen -s $save -e $eval_ -tw $tot_workers -wi $k  > /dev/null 2>&1 &

for x in "${ps[@]}"
do
    com="python  <path to file> -rdb -bs $batch_size -i $input_dir  -pb $pb -sb $sb -o $out_dir -fl $flen -s $save -e $eval_ -tw $tot_workers -wi $k  > /dev/null 2>&1 &"
    ssh -o "StrictHostKeyChecking no"  hadoop@$x $com
    echo $com
done

k=$((k+1))

for x in "${host[@]}"
do
    com="python  <path to file> -rdb -i $input_dir  -bs $batch_size  -pb $pb -sb $sb -o $out_dir -fl $flen -s $save -e $eval_ -tw $tot_workers -wi $k  > /dev/null 2>&1 &"
    ssh -o "StrictHostKeyChecking no"  hadoop@$x $com
    k=$((k+1))
done