Hi,

I'm currently using cp to copy files over night to a backup server using:

cp -pru /mnt/server-01/documents /home/backups/

This only copies files if they are newer than the destination (or if the destination file doesn't exist). What this won't protect me against is if someone overwrites a file with a blank one. The blank document will be newer than the old so it will overwrite the backup copy.

Is there a way to check if the file size is smaller, and if so create a different copy for example:

/home/backups/documents/me/quotes/computers.odt (100kb)
/home/backups/documents/me/quotes/computers-2.odt (94kb)

I hope that makes sense, thanks!

P.S. If cp cannot do this, what can?

You could use the -size option of the find cmd http://www.computerhope.com/unix/ufind.htm to check the size, passing in the size of the src file.
Might be overkill for each file.
The easiest might be to generate sorted lists of src & dest dirs, then compare same named file sizes (is use ls -l).
Basically, I think you need a short script.

Thanks for your suggestions. As there is a huge amount of data, running a find may take far too long.

I found that with cp you can specify the backup switch (-b) which creates a backup of the old file. For example:

/home/backups/documents/me/quotes/computers.odt~ (100kb)
/home/backups/documents/me/quotes/computers.odt (94kb) (newer)

The older of the two files has the ~ symbol at the end of the file name. These are always the two newest files (the backup file gets updated at the same time).

I think this is the easiest method, but it would be nice if it was a bit more flexible (for example to store 3 copies of the file rather than two. I know you can have it number these files, but the hard drive will fill up very quickly as there's no limit to the number it increases to!).

Like a I said, a simple/quick soln is just to ls -l src and destn dirs, piped through sort, and compare file sizes.