Zerui Wang

Start here,
make a difference.

Last updated on: Jun.12.2022

© 2015-2023. All rights reserved.

Using Git to Manage System Configuration Files on Linux/MacOS

06 Mar 2017

In this post, I will show you how to use git to manage your system configuration files (also known as dot files) on a Linux or MacOS machine. Most of you may have the experience of recovering a previous configurations of your heavily customized system. You might be tired of doing that manually (copy and paste them in the right locations) from time to time. It’s gonna be much easier to manage it by using git. Another advantage of using git is that you can roll back easily and have multiple versions (e.g., one for MacOS and one for Ubuntu) on different branches. There are several ways for implementation in terms of whether the git working directory is separated or not and the way we pull configurations from remote repositories. In this post, we only talk about how to set up a separated git repository from the working tree (home directory) for better management.

Creating a Separate Git Repository

First of all, let’s create a folder under home directory.

cd ~
mkdir configs

Then, we initialize the git repository within the folder we just created.

cd ~/configs
git init

If we do git status right now without any further configurations, we will get something similar as follows.

On branch master

Initial commit

nothing to commit (create/copy files and use "git add" to track)

It shows that nothing to commit. That’s because the git regards the current folder (~/configs) as working directory by default. The working tree can be set specifically by core.worktree git configuration option. Execute the following command under ~/config.

git config core.worktree "../../"

../../ is the relative directory with respect to the .git directory within ~/config. By setting that option, we make the git to keep the repository within ‘~/config’, but track the file under ~/ (two levels above).

Let’s check the status again by git status.

On branch master

Initial commit

Untracked files:
  (use "git add <file>..." to include in what will be committed)

    # following untracked files will be similar

    ../.bash_history
    ../.bash_logout
    ../.bash_profile
    ../.bashrc
    ../.emacs
    ../.vimrc

nothing added to commit but untracked files present (use "git add" to track)

Things are getting familiar right? Now, you can add your configuration files. For example:

    git add ../.bashrc
    git add ../.bash_profile
    git add ../.emacs
    git add ../.vimrc
    # etc.

We can then commit the changes we made above.

    git commit -m "Initial commit"

In order to make the git working properly, we have to create a file name .git in ~/ with the following content.

gitdir: /home/USERNAME/configs/.git

Don’t forget to replace the USERNAME with yours.

Setting up the Remote Git Server

It is always good to store your configurations on the cloud (e.g., GitHub (be careful with your private data!) and GitLab). It’s very easy to set up a remote git server.

git remote add origin https://github.com/YOURACCOUNT/YOURREPOSITORY.git
git push -u origin master

Resetting your configurations on a new machine

When you are setting up a new system, you can pull your configurations from the remote git repository. Since we are separating the working directory from the git repository, we have to specify where is the working directory before checkout the content in the repository. Instead of doing a normal git clone, do it with an option as follows:

git clone --no-checkout https://github.com/YOURACCOUNT/YOURREPOSITORY.git

After cloning the repository locally, we have to let the git know where is the working directory.

cd ~/configs
git config core.worktree "../../"

Now, we are ready to checkout the content. Basically, what we want here is to put the configuration files into right positions and overwrite if they are already existing.

git reset --hard origin/master

Pulling back changes on the remote repository

Once you set up your repository on multiple machines, you may want to update all of them by pulling the latest commit on the remote repository.

method 1

Create a file name .git in ~/ with the content: gitdir: /home/USERNAME/configs/.git.

cd ~
echo "gitdir: /home/USERNAME/configs/.git" > .git

method 2

Set the GIT_DIR to .git temporally.

cd ~/configs
export GIT_DIR=".git"

Now, we are ready to pull.

cd ~/configs
git pull

Ignoring files by .gitignore or .git/info/exclude

If you are interested in what’s the difference between .gitignore and .git/info/exclude, please look into this stack overflow link.

It’s usually annoying to see a lot of untracked files when you check the status. So just use either way mentioned above to ignore files that you don’t want git to track.

Thanks for reading!

Related Posts