Configuration Management in Linux

You can't yourself a linux enthusiast if you've never run into the problem of sorting out your configuration files. All of your precious files like .bashrc, .zshrc, .vimrc, .emacs, .fonts, .Xdefaults are some of the ones that can take a lot of pain to customize from the ground up.

Like most people, I started out by always backing up my important config files. But the list soon got so huge that I started having trouble remebering which files to back up, and to do it after every change.

So, obviously, I moved on to the next obvious option, symlinks! I started to keep all my config files in a single folder and started to symlink them to the places that they belonged to. I kept this up for a few months (mostly because it didn't seem like any trouble) when I finally decided to change my operating system. And the problem hit me. I already had over 50 configuration files under version control and had no real way to put them in correct places in the new OS. These files had accumulated over time and to put them back at correct places was an endeavour I wasn't going to undertake.

Now I tried to look for a tool that would help me manage and restore dotfiles at will. A friend suggested using vcsh and mr. You can find the relevant article here. I would've tried that too but it seemed too complicated to be an elegant way to handle this problem. For one thing, I wanted a script that could be a part of my dotfiles and not an external dependency. So as you might've already guessed, I started to plan out my own way to manage my dotfiles. So without further ado, I present to you:

Config Management, the UNIX way:

First things first, here is the repository for my project. The configuration system is inspired by VIM's modelines, which are basically comments that can control VIM's settings separately for code files. So, something like

// vim: set nonumbers:

in a source code's top or bottom lines would turn off line numbering in VIM when opening that source file. This is elegant as every config file has some comment system (and isn't worth using if it doesn't) and file properties can be easily specified in this way for every file.

Let's look at the file structure.

    ├── backup
    │   ├── 2015-11-30_13-46-31
    │   └── ........
    ├── configs
    │   ├── nvimrc
    │   ├── Xdefaults
    │   └── ........
    └── current_state

During the initial setup, you are only concerned with, and the folder configs. An empty current_state file may be needed. Your dotfiles and config folders go inside configs folder. The files can have any name and do not need to be named the same as the final target. For eg. .vimrc when kept inside the configs folder can be named mysuperamazingvimconfig and it won't matter. You can also group files of similar purpose together as the files will be searched recursively.

Now how do the files know where to go? Here is the easy part. Inside the corrsponding file, you just mention where that files need to go. So, for your mysuperamazingvimconfig, you will have the following in your source:

" place ''~/.vimrc''

The first double quote is to start a comment and the rest is the needed command. Please note that the path is enclosed by a pair of two single quotes not a double quote. Also, this comment will be looked for in the top 3 and the bottom 3 lines of the source code.

Another alternative for the command is:

" place &~/.vimrc&

especially if the file has a special meaning for the single quote (like .Xdefaults).

What about the folders?

For the folder that need to be symlinked, you create a file named folder_config inside the folder and put the same command as above. If the folder itself is being symlinked, then it won't be checked further for symlinkable files/folders. The rest of the folders, obviously will still be. For example, if you have a folder named fonts in configs and you want to place it as ~/.fonts, you will make a file folder_config inside fonts folder with the content:

place ''~/.fonts''

and you are done. All you have to do to create/update the symlinks is run ./ or python2

Under the Hood:

A lot of cool stuff happens under under the hood to ensure that no data loss occurs and no redundant files are created when is executed. Your current symlink status is stored in current_status so that the next time you change a place ''<path>'', the old symlink is deleted to prevent redundancy. If you manually remove the file from configuration management and place the real file in the actual place, then it will take care not to touch it even if it was part of current_status. While adding a file to config system for first time, if the file at target path is not a symlink but a real file, it is copied to the backups folder under current date and time before being replaced by a symlink to prevent data loss.

I hope you enjoyed reading this article.

Consider sharing this article using the links below.

Pallav Agarwal Image
My name is Pallav Agarwal. I am an undergrad of the department of Computer Science and Engineering, Indian Institute of Technology Kanpur, India. I love experimenting with tech, and this blog is a way for me to give a little helping hand to other's who are like me (but don't know it yet).

I am ambitious, intelligent, competitve (sometimes too much), loyal and brutally honest. People I respect the most are teachers, which is partially why I myself like to teach too. Apart from programming, I also like travelling, adventure sports and trying new food items. If you like a post, have a query, or just want to chit-chat, let me know here