Script to go Into Directories and Find/Delete files

Shell Programming and Scripting
1

I have a task, I usually do manually, but with growing responsibilities I tend to forget to do this weekly, I want to write a script that automates this, but I cant seem to work it out in my head, I have the shell of it out, but need help, and you guys have helped me with EVERY problem I have posted, and have found help in others posts as well. Please see below shell code I have started. I want to go to the ORDATA which is the variable for /opt/data in this case directory, find the folders with most stuff in it and cd into each one and then cd into the backup folder and do a couple of those find commands.

How do I make it go out of the first directory and into the next and so on and so forth? I know i can use the while loop in some capacity, but not quite sure.

for example /opt/data/ has 4 directories /mpw, /docprep, /edfiles, /vgdata and under each of those directories is a /backup directory as well, I want to start at /opt/data, go into each the 4 directories, go into the backup directories and do those find and delete commands i came with until its went into all 4 backup directories.

Below is what I have so far.

#!/bin/ksh
#Author:Emmanuel Iroanya Jr
#Date:April 2nd, 2013
#Purpose: The purpose of this is to check data and rollout directory and keep file space optimal. 

if [ "$1x" = "x" ]
	then
	echo "You must supply an environment parameter to this script"
	echo "example:  $0 tb82"
	exit
fi

. /opt/origenate/$1/config/usrconfig.sh 	#Sources Environments Admin Config Variables

cd $ORDATA 								#Environments Data Storage Directory
du -sk */ | sort -n -r | head -n 10 |awk '{print $2}' |while read d
do 
	cd "${d}" 							
	cd backup
	find *.zip -type f -a -mtime +3 -exec rm {} \;
	find *.pdf -type f -a -mtime +3 -exec rm {} \;
	find *.CSV -type f -a -mtime +3 -exec rm {} \;
	find *.txt -type f -a -mtime +3 -exec rm {} \;
done
2

Hello,

Per our forum rules, all threads must have a descriptive subject text. For example, do not post questions with subjects like "Help Me!", "Urgent!!" or "Doubt". Post subjects like "Execution Problems with Cron" or "Help with Backup Shell Script".

The reason for this is that nearly 95% of all visitors to this site come here because they are referred by a search engine. In order for future searches on your post (with answers) to work well, the subject field must be something useful and related to the problem!

In addition, current forum users who are kind enough to answer questions should be able to understand the essence of your query at first glance.

So, as a benefit and courtesy to current and future knowledge seekers, please be careful with your subject text. You might receive a forum infraction if you don't pay attention to this.

Thank you.

The UNIX and Linux Forums

1 Like
3

My apologies, that makes sense, I have changed the title of the post to something more appropriate.

Also, I put comments in the code/query so people could understand my thought process, does it not come across as a clear thought?

4

A few changes to suggest:

#!/bin/ksh
#Author:Emmanuel Iroanya Jr
#Date:April 2nd, 2013
#Purpose: The purpose of this is to check data and rollout directory and keep file space optimal. 

if [ -z "$1" ] # -z is simply "if string is blank".
then
	echo "You must supply an environment parameter to this script"
	echo "example:  $0 tb82"
	exit
fi

. /opt/origenate/$1/config/usrconfig.sh 	#Sources Environments Admin Config Variables

cd $ORDATA 								#Environments Data Storage Directory
du -sk */ | sort -n -r | head -n 10 |awk '{print $2}' |while read d
do
        # find's first parameter is a path, not a wildcard
        # You give it wildcards with -name 'string'
        # You can group them with ( ) and or them with -o
        # Remove the echo once you've tested and are sure it does what you want.
	find ${d}/backup '(' -name '*.zip' -o -name '*.CSV' -o -name '*.txt' ')' \
                -type f -a -mtime +3 -exec echo rm {} \;
done
1 Like
5

Appreciate it alot, let me play with this a little bit, question, Can I feed this multiple directories, I remembered I also have a tmp directory and files under ordata in a different structure tree other than /backup that fills quickly, so can i add

find ${d}/tmp '('

etc.. after that first find? It makes sense to me that it could, just want confirmation.

6

Yes, we can.:smiley:
If your find statements are exactly the same otherwise, you can combine them to one

find ${d}/backup ${d}/tmp -type f -mtime +3 '(' -name '*.zip' -o -name '*.CSV' -o -name '*.txt' ')' -exec echo rm -f {} \;

I changed the order, so the more "expensive" name comparison is after the simple file type and age checks. -a (AND) is only for readablity - can be omitted.
Last but not least, this should be secured:

cd $ORDATA || exit #fatal - directory must exist!
1 Like