git-without-github.md

Git Without Github

Github is fantastic!

That said, there may be a time when you need to clone, push, and pull to a git repository that does not live on Github but instead it lives on a remote computer not set up like Github. This note explains how to accomplish this.

I am capturing this information here because almost every bit of documentation I could find online assumes that you are either using Github or Gitlab. There are not many guides that clearly describe how to use git to track changes between two computers without also using Github or Gitlab.

Requirements

Here are the things you will need:

• SSH access to a remote server - I will use "server" to describe this computer from now on
  • with git installed & configured with your user information
• access to a client computer - I am using a laptop, so I will call it "laptop" in this note
  • with git installed & configured with your user information
• experience using git basic commands like git add, git commit, etc.
  • this note is not intended to be a git tutorial

Some Things to Know

• A directory (folder) is not a repository until you first initialize it with git
  • After you initialize the directory it becomes a repository
• You can't just start syncing a repository back-and-forth between your server and your laptop
  • You must first export that existing repository into what is called a bare repository
  • You will clone this bare repository
• I will use an imaginary user named Alice in the demonstration steps
• Existing repositories on your laptop that synchronize with an online service like Github or Gitlab won't be affected
  • New connections between your server and laptop will function independently of the existing connections that you already have

Detailed Tasks

Here is a list of all the tasks we will perform. The tasks are explained with examples below this list.

• initialize a directory into a git repository
• clone the git-initialized repository on the server into a bare repository on the server
• clone the bare repository from the server to the laptop
• clone the bare repository from the server to the server (that is not a typo)

Initializing a Directory into a git Repository

Let's say you have a directory (folder) on your server filled with server configuration scripts that you want track with git. Your directory is named server-scripts and it lives in your home directory on the server: /home/alice/server-scripts.

We will start by initializing the server-scripts directory into a git repository.

On the server, initialize the directory into a git repository by first navigating into the server-scripts directory and then issuing the git init command

$ cd /home/alice/server-scripts
$ git init

# You will get several lines of information about how to change the\
name later on - which you can ignore for now - and one final line\
that says:

Initialized empty Git repository in /home/alice/server-scripts/.git/

When you issued the git init command inside the server-scripts folder it added a hidden directory, .git that contains files and subdirectories that git will use to track any changes you make from now on.

Task: Clone the git-initialized Repository into a Bare Repository

Now, we need to create a bare repository. A bare repository is a storage location to which and from which remote users can synchronize. The bare repository is considered the 'central storage location'.

On the server, next we move up one level in the directory tree.

$ pwd
/home/alice/server-scripts

$ cd ..   # move up one relative level in the file tree

$ pwd
/home/alice

After that we will take the newly initialized git repository server-scripts on the server and copy it into a bare repository named new-bare-repo-on-server.git on the server. The .git suffix on the end of the filename helps people recognize that it is a git repo and not just a regular directory.

NOTE: the git repository and the new bare git repository can have the same name, but guides that use the same names for two different objects suck because this can be confusing to people learning something for the first time (like me).

In this example we use these file names:

• server-scripts

• new-bare-repo-on-server.git

  $ git clone\
  --bare\                      # tells git the type of repo to clone into
  server-scripts\              # directory that we initialized above
  new-bare-repo-on-server.git  # new bare repository
  
  Cloning into bare repository 'new-bare-repo-on-server.git`...
  done.
  

Ok, now we are ready to start cloning the bare repository named new-bare-repo-on-server.git on our laptop computer.

I have some very important points to make about this new-bare-repo-on-server.git later on in this note. Please read on.

Task: Clone Bare Repository From Server to Laptop

The act of cloning the bare repository from the server to the laptop will download the git bare repository to the laptop and set the repository up to:

• pull changes from the server to the laptop
• push changes to the server from the laptop

This is possible because that .git directory in the repository will have the URL for the server.

On your laptop, navigate to the location on your laptop where you want the git repository to be located. Let's say that you want to put the cloned repository in a specific directory named git-tracked-files on your laptop:

$ cd /home/alice/git-tracked-files/

On your laptop issue the following command to clone the repository from the server to the laptop. Note: we will change the name of the resulting cloned repository one more time to help keep things straight.

$ git clone [email protected]:/home/alice/new-bare-repo-on-server.git server-scripts-laptop-copy
Enter passphrase for key alice_id_ed25519: # Enter your SSH key passphrase if it has one
remote: Enumerting objects: done.
remote: Counting objects: done.
remote: Compressing objects: done.
remote: Total 233, reused 0, pack-reused 0
Receiving objects: 100% (233/233), done.

Now you have the git repository cloned onto your laptop as server-scripts-laptop-copy.

Notice that the newly-cloned directory does not have the .git suffix on the filename.

On your laptop, you can now make changes to the files in the server-scripts-laptop-copy repository. Then you can use the git add, git commit, git push, and git pull commands to synchronize changes between your laptop and your server.

Important note: In our example, we took a directory on the server server-scripts, we initialized it as a git repository, and we cloned the git repository into a bare repository new-bare-repo-on-server.git. When we cloned new-bare-repo-on-server.git from our server to our laptop, the we renamed the resulting repository on the laptop to server-scripts-laptop-copy (no .git suffix). Any changes made on the laptop's server-scripts-laptop-copy repository that are pushed/pulled to the server will only affect the new-bare-repo-on-server.git bare repository on the server NOT the original server-scripts repository on the server.

This brings me to my final point: tracking changes on the server.

Since we cloned server-scripts on the server to a bare repository new-bare-repo-on-server.git on the server, we no longer have a connection between server-scripts and new-bare-repo-on-server.git both on the server. While I am certain there is a way to connect the existing server-scripts repository on the server with the new-bare-repo-on-server.git bare repository also on the server, I have not yet been able to figure it out.

In the meantime, you can just clone the new-bare-repo-on-server.git on the server to a new repository the server named server-scripts-on-server. Then you can push/pull from server-scripts-on-server to new-bare-repo-on-server.git (both on the server) which can also then be pushed/pulled from server-scripts-laptop-copy on the laptop.

On the server:

$ git clone script-bare-repo.git server-scripts-on-server

This will result in a new repository on your server named server-scripts-on-server (without .git suffix). You can push/pull any changes you make in the server-scripts-on-server repository on the server to the new-bare-repo-on-server.git repository on the server.

Final Recap

We now have the following files and capabilities:

• server: server-scripts
  • basis for new-bare-repo-on-server.git on the server
  • is no longer connected to server-scripts on the server
• server: new-bare-repo-on-server.git
  • which can be pulled from or pushed to from server-scripts-laptop-copy the laptop
  • which can be pulled from or pushed to from server-scripts-on-server the server
  • which can be pulled from or pushed to from any other computer that clones this bare repository
• server: server-scripts-on-server
  • can push/pull to new-bare-repo-on-server.git on the server
• laptop: server-scripts-laptop-copy
  • clone of new-bare-repo-on-server.git on the server
  • can push to or pull from script-bare-repo.git on the server