• Keep a backup of single files scattered around your file system,

• Add those files to a Git repository

• When they change, they get backed up

• Restore your files

• A remote (preferably non-public) server to push your repository to.

If you are backing up similar configurations from different computers, you may want to keep several versions in branches. If that is the case, just create the branch locally and check it out.

• Create an empty Git repository:

$ mkdir ~/my_backup
$ git init

• Add some files, mimicking their full path:

$ mkdir -p etc
$ cp /etc/hosts etc/
$ git add etc/hosts

• Set up a git repository on your remote server:

$ mkdir -p ~/gitpath/my_backup
$ cd ~/gitpath/my_backup
$ git init --bare

• Add your remote server as ‘origin’:

$ git remote add origin ssh://my.remote.com/gitpath/my_backup

• Check commit and push work:

$ git commit -m "I'm starting by backing up my /etc/hosts"
$ git push

$ crontab -e
   13    23     *     *     *     cd ~/my_backup && git-confident

You can store extra files in your backup repository (like text notes) adding them to a .gcignore file:

$ echo SETUP_NOTES >> .gcignore
$ git add .gcignore

so git-confident won’t restore them. Similarly to .gitignore behaviour, you can ignore an entire folder by adding a .gcignore file inside it.

git-confident keeps a copy of each file added to git. This is not very useful when you have an entire folder to backup. You can track a whole folder using a .gcrecusrive file (acts like .gitignore):

$ touch etc/apache2/.gcrecusive
$ git add etc/apache2/.gcrecusive

• Source code

• Documentation

• Wiki

• Bug tracker

• Gem

• rsync

• eigenclass.org/hiki/gibak-0.3.0

• gitback - very similar to this project, but sets up the repository from a config file.

• Desktop backup systems

• Time machine (on OS X)