Bash script to clean files in directory based on access time and created time. https://labbots.com/delete-stale-files-by-access-time-bash/

file_cache_clean.sh

#!/bin/bash
#!/bin/bashset -e

# Script to cleanup cache files in file system.
# This script is used to make sure the cache folder does not exceed the set threshold limit in the server.
# The script can be configured as cron job in the server to monitor and delete older files in specified directory.
# The script sends out email if the script cannot bring the directory size below the threshold even after deleting
# files based on min and max access time and created time set in the script.
#
# Usage
#------
# ./file_cache_clean.sh <directory> <threshold_in_mb>
#
# Note: Make sure the threshold is set in MB and if not provided it defaults to 100MB
# The script is required to be executed as root.
#
# Caution : Make sure the directory argument is correct. Since the script is executed as root. Any wrong directory input can 
# corrupt the system.

# Incase of failure to minimize to threshold notify
EMailTo="[email protected]"
DateStamp=$(date '+%Y%m%d_%H%M')

# Global varible declarations
FOLDER="$1"
THRESHOLD=${2:-100}
VERBOSE=true
MAX_ACCESSTIME_LIMIT=30
MIN_ACCESSTIME_LIMIT=10
MAX_CREATETIME_LIMIT=30
MIN_CREATETIME_LIMIT=10

function log() {
	error="$1"
	if [[ "$VERBOSE" = true ]] || [[ "$error" = true ]] ; then
		echo -e "${1}"
	fi
}

# Check whether the script is run as root
if [[ $UID -ne 0 ]];then
	log "Script needs to be run as root.Aborted with no further action." true
    exit 1
fi

# Validation of input
if [[ "$FOLDER" == "/" ]];then
	log "Forbidden." true
	exit 1
fi

if [[ -z "$FOLDER" ]];then
	log "Folder argument is mandatory." true
	exit 1
fi

if [[ ! -d "$FOLDER" ]];then
	log "Specified Folder does not exist." true
	exit 1
else
	# Resolve relative path or symlinks to actual path
	FOLDER="$(readlink -f $FOLDER)"
fi
# End of Validation

# Delete all files in directory based on access time
function cleanUpDirectoryByAccessTime(){
	ACCESSTIME=$1
	FOLDERSIZE=$(du -ms $FOLDER | awk '{print $1}')

	if [[ "$FOLDERSIZE" -gt "$THRESHOLD" ]] ;then
		find "$FOLDER" -type f -atime +${ACCESSTIME} -delete
		FOLDERSIZE=$(du -ms $FOLDER | awk '{print $1}')
		if [[ "$FOLDERSIZE" -gt "$THRESHOLD" ]];then
			access_dcr=$(($ACCESSTIME-10))
			if [[ "$access_dcr" -ge "$MIN_ACCESSTIME_LIMIT" ]];then
				cleanUpDirectoryByAccessTime $access_dcr
			fi
		fi
	else
		return 0;
	fi
} 

# Delete all files in directory based on create time
function cleanUpDirectoryByCreatedTime(){
	CREATETIME=$1
	FOLDERSIZE=$(du -ms $FOLDER | awk '{print $1}')
	
	if [[ "$FOLDERSIZE" -gt "$THRESHOLD" ]];then
		find "$FOLDER" -type f -ctime +${CREATETIME} -delete
		FOLDERSIZE=$(du -ms $FOLDER | awk '{print $1}')
		if [[ "$FOLDERSIZE" -gt "$THRESHOLD" ]];then
			access_dcr=$(($CREATETIME-10))
			if [[ "$access_dcr" -ge "$MIN_CREATETIME_LIMIT" ]];then
				cleanUpDirectoryByCreatedTime $access_dcr
			fi
		fi
	else
		return 0;
	fi
}

log "Starting to process $FOLDER for cleanup by access time."
cleanUpDirectoryByAccessTime $MAX_ACCESSTIME_LIMIT

log "Starting to process $FOLDER for cleanup by created time."
cleanUpDirectoryByCreatedTime $MAX_CREATETIME_LIMIT

log "Delete all empty folders."
find "$FOLDER" -empty -type d -delete

FOLDERSIZE=$(du -ms $FOLDER | awk '{print $1}')
if [[ "$FOLDERSIZE" -gt "$THRESHOLD" ]];then
	echo "folder $FOLDER is above threshold and script cannot clean up the directory.Date: $DateStamp" | mail -s "File cache clearout" ${EMailTo} > /dev/null 2>&1
fi