Remove old files from Hadoop tmp

hadoop_remove_old_tmp.sh

today=`date +'%s'` # date today
files=`hdfs dfs -ls /tmp | tail -n +2` # all files in tmp
granularity=$(( 24*60*60 )) # granularity of time. Now set to days
olderThan=7 # granularity times olderThan gives you what age files should be deleted

for line in $files; do
  dir_date=$(echo ${line} | awk '{print $6}')
  # difference=$(( ( ${today} - $(date -j -u -f "%Y-%m-%d %H:%M" ${dir_date} +%s) ) / ${granularity} )) # MacOS
  difference=$(( ( ${today} - $(date -d ${dir_date} +%s) ) / ${granularity} )) # Linux
  filePath=$(echo ${line} | awk '{print $8}')

  if ([ "${difference}" -gt $olderThan ]); then
    # hadoop fs -rm -r ${filePath}
    echo $filePath
  fi
done