How to diff two folders by inodes

Question

I'm currently using a script to backup files on my computer.

In a nutshell, it works like this:

rsync --link-dest=$FOLDER/current $SOURCE $DESTINATION

However, I think that my script may be duplicating some files rather than hard linking them. I would like to verify which files are (or aren't) detected as duplicates. I know that the file has been hardlinked correctly of the two files share the same inode.

How can I do a diff of two folders by inodes?

Answer 1

#!/bin/sh
cd path1
find . -type f -exec ls -i {} + | sort -k2 > ~/path1.lst
cd path2
find . -type f -exec ls -i {} + | sort -k2 > ~/path2.lst
cd ~
diff -U0 path1.lst path2.lst

Answer 2

If i understand correctly, your backup files are stored with duplicates. I think you can add to script date stamp or/and checksum, like md5sum. For example:

DESTINATION="(date +%F-%H-%M).foo-files"

#!/bin/sh
### Checking md5 hash ###

x='foo'
y='bar'
z=`echo $x | /usr/bin/md5sum`
a=`echo $y | /usr/bin/md5sum`
if [ "$z" = "$a" ];then
        echo '=-=-=-Stuff_duplicate-=-=-='
        . . .
fi

Notice: MD5-hash will changed if only one or more symbols are differents in two files.

Good luck! =)

Answer 3

Not sure if I understand correctly what you mean but when you mention hardlink I hope you mean hardlink between the backups and not original and backup files.

You can find which files have the same inode with find.

find dest_dir -samefile orig_dir/file

this command will provide you with a file list that matches the same inode as the original one.

If you will put this inside a loop you can do something similar to

for i in `ls orig_dir`
do
        result=`find dir2/ -samefile dir1/firstfile`
        if [ result  != "" ]
        then
                echo "Duplicate: " $i
        fi
done

This will print all the files that are hardlink between original directory and destination directory by inode. Notice that I'm performing the if based on the output as find always returns true using this flags. from the man page.

All options always return true. Except for -daystart, -follow and -regextype