#! /usr/bin/env bash

LOCAL="/user/hjudge/"
FILENAME="testFile"
dd if=/dev/zero of=$FILENAME bs=$1 count=1
hdfs dfs -Ddfs.replication=1 -put $FILENAME
hdfs fsck $LOCAL$FILENAME -files -blocks -locations | grep -i 'blk_' \
| while read line;
do
  ID="$(echo "$line" | awk '{print $2;}' \
  | awk '{print substr($0,index($0,":") + 1);}' | awk '{print substr($0,index($0,"_") + 1);}' \
  | awk '{print "blk_"substr($0,0,index($0,"_") - 1);}')";
  NODE="$(hdfs fsck -blockId $ID | grep -i "^Block replica" | head -n 1 | awk '{print substr($5,0,index($5,"/") - 1);}')";
  echo $NODE
  sudo -u hdfsuser ssh hdfsuser@$NODE "cd /dfs/; find -name $ID && exit" < /dev/null
done > file
SUM=0
while read node;
do
  read path
  location="$(echo $path | awk '{print "/dfs"substr($0,2)}')"
  NUM="$(sudo -u hdfsuser ssh hdfsuser@$node "ls -l $location && exit" < /dev/null | awk '{print $5}')"
  SUM="$(echo $SUM $NUM | awk '{$1+=$2; print $1}')"
done < file
echo $SUM $1 | awk '{$2-=$1; print $2}'
rm file
hdfs dfs -rm $FILENAME
rm $FILENAME