This script was written as a result of the space that metche backups were taking on wiki:PuffinServer, see ticket:531.

#!/bin/bash

# This script can find or delete files in $DIR older than $DAYS
# which are not symlinks and are not linked to by symlinks in 
# $DIR, it has been written to be run via cron to prevent metche 
# using too much disk space.

# Number of days worth of files to save
DAYS=1

# Directory where the files are
DIR="/var/lib/metche"

# check that the script is being run by root
if [[ "$(id -u)" != "0" ]] ; then
  echo "You must run $0 as root or via sudo" 
  exit 2
fi

# Check that $DIR exists
if [[ ! -d $DIR ]]; then
  echo "The directory $DIR doesn't exist, please edit the DIR variable in $0 and try again" 
  exit 2
fi

# Symlinks in $DIR 
SYMLINKS=$(find $DIR -maxdepth 1 -type l)

# Destinations of the symlinks | seperated
for symlink in $SYMLINKS; do
  # Get the destination of the symlink 
  DESTPATH=$(readlink $symlink)
  # Get the symlink filename
  DESTFILE=$(basename $DESTPATH)
  # If $SYMDEST doesn't exist then create it using the value of $DEST
  # and if it does exist then add $DEST to it
  if [[ ! $SYMDESTS ]]; then
    SYMDESTS="$DESTFILE"
  else
    SYMDESTS="$SYMDESTS|$DESTFILE"
  fi
done

# Change the space seperated list of symlinks to a | seperated one
for symlink in $SYMLINKS; do
  LINK=$(basename $symlink)
  # If $LINKS doesn't exist then create it using the value of $LINK
  # and if it does exist then add $LINK to it
  if [[ ! $LINKS ]]; then
    LINKS="$LINK"
  else
    LINKS="$LINKS|$LINK"
  fi
done

# The symlinks and their destinations which we want to keep
KEEPFILES=$(echo "$LINKS|$SYMDESTS")

# Files older than $DAYS in $DIR
OLDFILES=$(find $DIR -mtime +$DAYS -type f)

# Delete files which are older than $DAYS and which 
# are not listed in $KEEPFILES 
if [[ $OLDFILES ]]; then
  # Loop through the $OLDFILES
  for file in $OLDFILES; do
    # Remove the $DIR from $file
    FILE=$(basename $file)
    # $MATCH only has a value if $FILE is not in $KEEPFILES
    MATCH=$(echo $FILE | egrep -v "^($KEEPFILES)$")
    if [[ $MATCH ]]; then
      # Check if there is a -d on the command line
      if [[ $1 == "-d" ]]; then
        # Delete the files if it's not in the $KEEPFILES list 
        echo "Deleting $DIR/$FILE"
        rm $DIR/$FILE
      else
        echo "Deletable file older than $DAYS days: $DIR/$FILE"
      fi
    fi
  done
else
  echo "No deletable files older that $DAYS days found in $DIR" 
fi

# Check if there is a -d on the command line
if [[ $1 != "-d" ]]; then
  # We are not going to delete anything
  echo "To remove deletable files run: $0 -d"
fi