MapR checkcomp

checkcomp.sh

#!/bin/bash
# checkcomp - A script to show the relative compressed and uncompressd file sizes on a MapR filesystem.
# Chris Matta
# cmatta@mapr.com
#
# Currently broken when a directory has multiple dir children. Need to write a directory walk funciton.

set -o nounset
set -o errexit

SCRIPTNAME=$0;
REPORT_GB=0;
HADOOP=/usr/bin/hadoop;


usage () {
  echo "Usage: ${SCRIPTNAME}: [-h] <file name or directory name of MapR FS>";
  exit 1;
}

getUncompressed () {
  echo $($HADOOP mfs -lss $1 | grep -v "^\s" | awk '/\// {total = total + $7}END{print total}')
}

getCompressed () {
  blocks=$($HADOOP mfs -lss $1 | grep -v "^\s" | awk '/\// {total = total + $8}END{print total}')
  echo $(($blocks * 8192));
}

humanReadableSize () {
  # Return the human readable size for a file size given in bytes
  suffix="b";
  size=$1;
  if [[ $1 -ge 1024 ]]
  then
    suffix="K";
    size=$(($size/1024));
  fi

  if [[ $1 -ge $((1024 * 1024)) ]]
  then
    suffix="M";
    size=$(($size/1024));
  fi

  if [[ $1 -ge $((1024 * 1024 * 1024)) ]]
  then
    suffix="G";
    size=$(($size/1024));
  fi

  echo "${size}${suffix}";
}

if [[ $# -eq 0 ]]
then
  usage
fi

if [[ $# -eq 1 ]]
then
  FILENAME=$1;
  echo $FILENAME
fi

if [[ $# -eq 2 ]]
then
  if [[ $1 = "-h" ]]
  then
    FILENAME=$2;
    REPORT_GB=1;
  else
    usage
  fi
fi

compressed=$(getCompressed $FILENAME);
uncompressed=$(getUncompressed $FILENAME);

if [[ $REPORT_GB -eq 1 ]]
then
  echo "$(humanReadableSize $compressed) compressed";
  echo "$(humanReadableSize $uncompressed) uncompressed";
  echo  $(humanReadableSize $compressed) / $(humanReadableSize $uncompressed)\($(echo "scale = 2; $compressed * 100 / $uncompressed" | bc)%\)
else
  echo "$compressed compressed";
  echo "$uncompressed uncompressed";
  echo $compressed / $uncompressed \($(echo "scale = 2; $compressed * 100 / $uncompressed" | bc)%\)
fi