Copy of script
#!/bin/bash
#
# SCRIPT: saw-retention.sh
# AUTHOR: David Waugh
# DATE: 18/02/2015
# REV: 1.0a
#
# PLATFORM: Linux Centos (Security Analytics 10.X)
#
# PURPOSE: Calculate Data Retention of a Warehouse
#
#
######################################################################
#
# Expects the path to the Warehouse as the first argument
# eg the argument given should point to the directory ABOVE rsasoc
# My files are written to /mnt/saw/lonsaw/rsasoc/v1/logs/data/2015/2
# So I will give /mnt/saw/lonsaw as the first argument
# We will calculate
# MIN DAILY USAGE RATE
# MAX DAILY USAGE RATE
# AVG DAILY USAGE RATE
MIN_DAILY=10000000000
MAX_DAILY=0
AVG_DAILY=0
DAY_COUNT=0
DAYSIZE=0
TOTALSIZE=0
echo -e "*** Looking at Warehouse under Mount $1 ***\n"
#First we do Logs
for years in $(ls -1 $1/rsasoc/v1/logs/data)
do
#echo -e "Year: $years\n"
for month in $(ls -1 $1/rsasoc/v1/logs/data/$years)
do
#echo -e "Month: $month\n"
for day in $(ls -1 $1/rsasoc/v1/logs/data/$years/$month)
do
#echo -e "Day: $day "
DAYLOGSIZE=`du -k $1/rsasoc/v1/logs/data/$years/$month/$day --max-depth=0 |tail -n 1 |cut -f 1`
DAYSESSIONSIZE=`du -k $1/rsasoc/v1/sessions/data/$years/$month/$day --max-depth=0 |tail -n 1 |cut -f 1`
DAYSIZE=`expr $DAYLOGSIZE + $DAYSESSIONSIZE`
TOTALSIZE=`expr $TOTALSIZE + $DAYSIZE`
echo -e "$day/$month/$years LOGS: $DAYLOGSIZE SESSIONS: $DAYSESSIONSIZE TOTAL: $DAYSIZE "
# Only Increment Day Count for Non Empty Directories
if [ $DAYSIZE -gt 0 ]
then
DAY_COUNT=`expr $DAY_COUNT + 1`
fi
if [ $DAYSIZE -lt $MIN_DAILY ]
then
MIN_DAILY=$DAYSIZE
fi
if [ $DAYSIZE -gt $MAX_DAILY ]
then
MAX_DAILY=$DAYSIZE
fi
done
done
done
AVG_DAILY=`expr $TOTALSIZE / $DAY_COUNT`
FREESPACE=`df $1 | tail -n 1 |tr -s ' ' |cut -d ' ' -f4`
DAYSLEFT_AVG=`expr $FREESPACE / $AVG_DAILY`
DAYSLEFT_MAX=`expr $FREESPACE / $MAX_DAILY`
DAYSLEFT_2MAX=`expr $FREESPACE / $MAX_DAILY / 2`
# Convert to MB
MIN_DAILY=`expr $MIN_DAILY / 1024`
MAX_DAILY=`expr $MAX_DAILY / 1024`
AVG_DAILY=`expr $AVG_DAILY / 1024`
TWICEMAX=`expr 2 \* $MAX_DAILY`
echo -e "MIN(MB): $MIN_DAILY MAX(MB): $MAX_DAILY AVG(MB): $AVG_DAILY DAYS: $DAY_COUNT \n"
echo -e "DAYS LEFT AT AVERAGE: $AVG_DAILY MB/ DAY: $DAYSLEFT_AVG"
echo -e "DAYS LEFT AT MAXIMUM: $MAX_DAILY MB/ DAY: $DAYSLEFT_MAX"
echo -e "DAYS LEFT AT TWICE MAXIMUM $TWICEMAX MB /DAY: $DAYSLEFT_2MAX" | 
on Jun 14, 2016