This website and its contents are deployed to my VPS via git. The repository is pushed to a production remote and uses hooks to run some basic scripts that organize the incoming content. This is almost CI/CD, but not quite; regardless, git can be a powerful tool to build simple deployment pipelines to remote servers.
Local and Remote Repo Setup
First, we want to set up the basic development repository with a git init. I always have a remote backup to push to on Codeberg, weather I plan on publishing the code publically or not, for some redundancy. Run a git remote add origin git@example.com/me/my-remote, and of course replace this example repo URL with your real one.
Next, we're going to set up our production remote. Connect to your target server, and navigate to the /var/www directory, then run mkdir my-website-bare. In this post, I'm going to be using a static website as an example, and it's common for production websites to live in this directory. Once, there run git init --bare to initialize a bare git repository. You'll notice that the bare repo does not have a .git directory, but instead will show all the contents of what would be in that directory at the root.
❯ ls -l
total 12
drwxr-xr-x. 1 brandon brandon 0 Apr 30 15:11 branches
-rw-r--r--. 1 brandon brandon 66 Apr 30 15:11 config
-rw-r--r--. 1 brandon brandon 73 Apr 30 15:11 description
-rw-r--r--. 1 brandon brandon 21 Apr 30 15:11 HEAD
drwxr-xr-x. 1 brandon brandon 506 Apr 30 15:11 hooks
drwxr-xr-x. 1 brandon brandon 14 Apr 30 15:11 info
drwxr-xr-x. 1 brandon brandon 16 Apr 30 15:11 objects
drwxr-xr-x. 1 brandon brandon 18 Apr 30 15:11 refs
Now, let's add this bare repository as a remote to our local repo by running the following:
git remote add prod me@myserver:/var/www/my-website-bare
For the purposes of this little walk through, I won't get into the weeds of each of these files and directories; the only directory we're concerned with here is hooks. Git hooks are scripts that Git will execute when certain events happen in the repository. They are flexible and very useful when automating workflows that involve Git. I use them in my website's production workflow, and in some of my development workflows.
The post-receive Hook
The workflow we're aiming for here is as follows:
- Push to the production repository from your local repo.
- When the prod remote receives the new code it will distribute it where it needs to be.
- Once the code has been updated, we can also rebuild the static site or run any commands needed to update the website with the changes.
In your bare repo, create a file called post-receive in the hooks directory, and write the following in it:
#!/bin/bash
# Unset the git environment variables
local_desc=$(git describe)
foreign_desc=$(unset $(git rev-parse --local-env-vars); git -C /var/www/my-website-prod/.git describe)
# navigate to the production repo and execute
# a git pull on the prod main branch
cd /var/www/my-website-prod || exit
git --git-dir=/var/www/my-website-prod/.git pull prod main
The first thing you'll notice here, is we have another website repository at /var/www/my-website-prod. This is where the website will actually live, and where your web server will be pointing to your static files. Go ahead and navigate there and run the following:
git clone /var/www/my-website-bare /var/www/my-website-prod
This will clone your bare repo into a new, un-bare repo located in my-website-prod.
So to break down our post-receive hook: Lines 4, 5, and 6 will unset the git environment variables from our bare repository, and then set them to the env variables from our production repository at /var/www/my-website-prod. Next, we will navigate to the prod repo and execute a
git pull prod main, to pull the main branch of the production remote to our production repository, deploying the code.
Now, when you run a git push prod main in your local repository, it will push to the bare remote, the post-receive hook will trigger, and will then pull the code from the bare repo, to the production repo.
Conclusion
The flexibility of Git hooks are very useful. I throw in other script calls to run tests and rebuild my website on a single push, allowing me to deploy changes to the website without ever having to ssh to my production server, creating a small and efficient CI/CD pipeline for my
personal website. Easy peasy, right? Have fun!
Top comments (0)