git & git-lfs

Overview

Git is a version control system, which allows you to keep old versions of files and directories (usually source code), keep a log of who, when, and why changes occurred, etc.,

From June 2025, version 2.43.5 is the system-wide default (if you load no modulefile.) Hence it is no longer necessary to load a modulefile, unless you specifically need an older version!

git-lfs Git Large File Storage (LFS) replaces large files such as audio samples, videos, datasets, and graphics with text pointers inside Git, while storing the file contents on a remote server like GitHub.com or GitHub Enterprise.

If you require a different version to the one installed we advise that you install it in your home directory.

Restrictions on use

All users of the system can access the software.

Set up procedure

We recommend using the newer, system-wide version of git by NOT loading any modulefiles. This gives you immediate access to version 2.43.5 (note – if we install newer versions of git, we will revert back to providing them via modulefiles.)

# No modulefile loaded
you have access to git 2.43.5

# Or load one of the following modulefiles
module load tools/gcc/git/2.43.0
module load tools/gcc/git/2.24.0

To use git-lfs use:

module load tools/bintools/git-lfs/2.8.0

Running the application

On the command line type git.

If you wish to prepare a repository for use with git-lfs run the following inside each repository:

git lfs install

External Access using https://

You can run your commands directly on the login node, or on a compute node (via srun), or in a batch job. For example:

From the login node or in a jobscript:

git clone https://github.com/myproject.git

Or, from an interactive session on a compute node:

# On the login node, request an interactive session
srun -p interactive -t 0-1 --pty bash
  #
  # Wait until you are logged in to a compute node

# Now run your commands as normal
git clone https://github.com/myproject.git

Note: If you get an error about the web-proxy, please edit your ~/.gitconfig file:

gedit ~/.gitconfig
  #
  # Remove 'proxy' line from the [http] section
  [http]
  proxy = http://some.proxy.addr:3128/       # REMOVE THIS LINE FROM YOUR ~/.gitconfig FILE

Note: You may also need to remove some old config from your ~/.ssh/config file if using the ssh:// protocol (see below)

Once cloned, you may wish to work with a specific tag:

# List all tags (and those matching a pattern)
git tag
git tag --list v2*

# Work with a specific tag
git checkout tagname

External Access to github.com using https:// (with Personal Access Tokens)

Github are moving away from using https:// with passwords to access your repos. Their blog post says:

If you are using a password to authenticate Git operations with GitHub.com today, you must begin using a personal access token over HTTPS or SSH key by August 13, 2021, to avoid disruption.

See their documentation on how to create a Personal Access Token (this is similar to a random, expiring password.)

External Access to github.com using ssh:// (with key)

An alternative to using Personal Access Tokens with https:// is to set up a public-private key-pair for your github.com account, which will give you password-less access to github.com via ssh://.

The following is a quick summary of how to setup ssh:// access to github on the CSF using a key-pair:

# If not already done so, create a key-pair on the CSF:
ssh-keygen -t ed25519 -C your.email@manchester.ac.uk.com -f ~/.ssh/id_github
  #
  # We recommend protecting your key with a passphrase when asked, but this is optional.
  # If you intend to run git from batch jobs, do not set a passphrase.

# Now edit your ~/.ssh/config file and add the following:
gedit ~/.ssh/config

# ADD the following config to make ssh:// use your key
Host github.com
  IdentityFile ~/.ssh/id_github
  IdentitiesOnly yes

  # REMOVE any old config if present
  # HostName nataas.itservices.manchester.ac.uk
  # IdentityFile /mnt/iusers01/support/mbexegc2/.ssh/id_ed25519
  # StrictHostKeyChecking  no
  # Port 2022

(save the file)

# Now look at the public key-file contents (we'll need the displayed line shortly):
cat .ssh/id_github.pub
ssh-ed25519 AAA................... your.email@manchester.ac.uk

Now login to your github.com account as normal.

1. Then go to your Account Settings
2. Then SSH and GPG Keys.
3. Press the New SSH Key green button and paste in the above line.

Finally, test you connection from the CSF:

# On the CSF login node:
# Use the github test method to check your keys
ssh -T git@github.com
  # If you saved your key with a passphrase you'll be prompted to enter it
# You should see
Hi githubusername! You've successfully authenticated, but GitHub does not provide shell access.

# Return to the login node
exit

Don’t clone in to scratch

We strongly recommend that you clone repositories in to you home area, or other research data storage, instead of your scratch area. This is because the scratch clean-up policy can automatically delete files not accessed for at least 3 months. A cloned repository often contains many files that you might not access and so the scratch cleanup could remove files without you being aware of it.

VSCode and github.com

Some VSCode extensions will attempt to contact github.com from the CSF login node. This occurs when you have VSCode installed on your local PC/laptop, and you use the “SSH Extension” in VSCode to login to the CSF and edit code in your CSF home directory. But it is not possible to make VSCode run the git commands on a CSF compute node, as is normally required.

To allow VSCode to work with github.com, some access to github from the login node will now work. We have added github.com to the “allow-list” of sites that can be contacted from the login node. Please note that the Research Infrastructure team DO NOT manage the list of permitted sites that can be contacted from the CSF login nodes. We can only make a request that access be granted to certain sites.

If your use of VSCode is trying to contact a github service from the CSF login node and is being blocked, please contact us with the domain name (e.g., something.github.com) that is reported in any error message you receive.

Further info

• On the login node run: git help
• http://git-scm.com/

Updates

None