Hi,

I have a directory which contains in it multiple files (backups of virtual machines). They are named (for example):

webserver-19012011.xva
webserver-12012011.xva
webserver-05012011.xva
dbserver-19012011.xva
dbserver-12012011.xva
dbserver-05012011.xva

What I need to do is commit a diff (the files are in excess of 8GB) of the two latest files in each set (webserver diff and db server diff) and if they are identical delete one of them (not bothered which). The modification dates match the last characters, so ls -t is an option.

I've been working on this logic, however given my example it would only check the two webserver VM's:

I'm struggling with how to pair them up for the diff more than anything else :S. I think I can probably handle the rm bit, however it maybe easier to have a diff of two variables and then delete one of them.

Thanks, sorry if this is too verbose!

The diff command is designed for line oriented text files. Don't expect it to be able to handle binary files. To compare two binary files use `cmp'. It will tell you the offset where the files differ or they if are the same.

cmp file1 file2 && rm file1

1 members found this post helpful.

Wow, thanks for that! I didn't even know I was looking through the diff man page and remember reading the word binary somewhere so assumed it would function as expected.

Evidently I've not tested this enough, diff and cmp both say the files differ. cmp says they differ on the first line; this is just a long string of settings used when commiting an import and isn't relevant to the actual data...

Just tried this with -l, thousands of differences; joy oh joy, new diff method needed I suspect, or perhaps an entirely new idea.

Thanks for your help guys