Tag: rm

The accomplishment for the week (Tech!)

So where I work – instead of doing standard mysqldump SQL dumps – they copy the INNODB files into a backup. These backups an beĀ  6GB or more per week. So it was killing the VM the MySQL server runs on.

Now my predecessor had started writing a prune-mysql-backups.sh script. All he’d done was do a list files (ls -At) in time order and then create a text file called filelist.

But the thing was, the dated files went in ascending order. I didn’t want them that way. And apparently he didn’t know enough about sort and sed to complete the script. Sort or sort does exactly what it says, sorts a list of items. The term sed means Stream Editor.

So I had to fire sort -r which means in reverse chronological order then I ran sed to strip out the three newer items in the list, and then pass that file in to a section that walked through the list and rm -rf’d the directories.

That last part ‘rm -rf’ is a goodie. In Unix parlance rm means REMOVE and the -r means recursive the f being all files. You never, ever, ever want to do this on the root partition which would be ‘rm -rf /’. A fast and hard rule to live by is something has to live in FRONT of the slash, e.g. ‘rm -rf 2013-09-30_21-00-34/ ‘would be ok since it’s going to try to walk in that directory and then delete the content and directory.

Another trick is to alias rm -rf and have it make a confirmation of “Do you really want to do this?” before proceeding. That’s the power of Unix/Linux. You can write scripts and aliases to do pretty much anything.