Git Large File Storage (LFS)

Git Large File Storage (LFS) is an open source Git extension that helps Git repositories manage large binary files efficiently. Git can’t track changes to binary files (like audio, video, or image files) the same way it tracks changes to text files. While text-based files can generate plaintext diffs, any change to a binary file requires Git to completely replace the file in the repository. Repeated changes to large files increase your repository’s size. Over time, this increase in size can slow down regular Git operations like clone, fetch, or pull.

Use Git LFS to store large binary files outside of your Git repository, leaving only a small, text-based pointer for Git to manage. When you add a file to your repository using Git LFS, GitLab:

  1. Adds the file to your project’s configured object storage, instead of the Git repository.
  2. Adds a pointer to your Git repository, instead of the large file. The pointer contains information about your file, like this:

    version https://git-lfs.github.com/spec/v1
    oid sha256:lpca0iva5kpz9wva5rgsqsicxrxrkbjr0bh4sy6rz08g2c4tyc441rto5j5bctit
    size 804
    
    • Version - the version of the Git LFS specification in use
    • OID - The hashing method used, and a unique object ID, in the form {hash-method}:{hash}.
    • Size - The file size, in bytes.
  3. Queues a job to recalculate your project’s statistics, including storage size and LFS object storage. Your LFS object storage is the sum of the size of all LFS objects associated with your repository.

Files managed with Git LFS show a LFS badge next to the filename:

Git LFS tracking status

Git LFS clients use HTTP Basic authentication, and communicate with your server over HTTPS. After you authenticate the request, the Git LFS client receives instructions on where to fetch (or push) the large file.

Your Git repository remains smaller, which helps you adhere to repository size limits. For more information, see repository size limits for self-managed and for GitLab.com.

Understand how Git LFS works with forks

When you fork a repository, your fork includes the upstream repository’s existing LFS objects that existed at the time of your fork. If you add new LFS objects to your fork, they belong to only your fork, and not the upstream repository. The total object storage increases only for your fork.

When you create a merge request from your fork back to the upstream project, and your merge request contains a new Git LFS object, GitLab associates the new LFS object with the upstream project after merge.

Configure Git LFS for a project

Tier: Free, Premium, Ultimate Offering: Self-managed, GitLab Dedicated

GitLab enables Git LFS by default for both self-managed instances and GitLab.com. It offers both server settings and project-specific settings.

  • To configure Git LFS on your instance, such as setting up remote object storage, see GitLab Git Large File Storage (LFS) Administration.
  • To configure Git LFS for a specific project:

    1. In the root directory of your local copy of the repository, run git lfs install. This command adds:
      • A pre-push Git hook to your repository.
      • A .gitattributes file to track handling for individual files and file types.
    2. Add the files and file types you want to track with Git LFS.

Add a file with Git LFS

Prerequisites:

To add a large file into your Git repository and immediately track it with Git LFS:

  1. To track all files of a certain type with Git LFS, rather than a single file, run this command, replacing iso with your desired file type:

    git lfs track "*.iso"
    

    This command creates a .gitattributes file with instructions to handle all ISO files with Git LFS. The line in your .gitattributes file looks like this:

    *.iso filter=lfs -text
    
  2. Add a file of that type (.iso) to your repository.
  3. Tell Git to track the changes to both the .gitattributes file and the .iso file:

    git add .
    
  4. To ensure you’ve added both files, run git status. If the .gitattributes file isn’t included in your commit, users who clone your repository don’t get the files they need.
  5. Commit both files to your local copy of your repository:

    git commit -am "Add an ISO file and .gitattributes"
    
  6. Push your changes back upstream, replacing main with the name of your branch:

    git push origin main
    

    Make sure the files you are changing aren’t listed in a .gitignore file. If this file (or file type) is in your .gitignore file, Git commits the change locally, but does not push it to your upstream repository.

  7. Create your merge request.

Add a file type to Git LFS

When you add a new file type into Git LFS tracking, existing files of this type are not converted to Git LFS. Files of this type added after you begin tracking are added to Git LFS. To convert existing files of that type to use Git LFS, use git lfs migrate.

Prerequisites:

To start tracking a file type in Git LFS:

  1. Make sure this file type isn’t listed in your project’s .gitignore file. If this file type is in your .gitignore file, Git commits your changes locally, but does not push it to your upstream repository.
  2. Decide what file types to track with Git LFS. For each file type, run this command, replacing iso with your desired file type:

    git lfs track "*.iso"
    
  3. Tell Git to track the changes to the .gitattributes file. Commit the file to your local copy of your repository, replacing iso with your desired file type:

    git add .
    git commit -am "Use Git LFS for files of type .iso"
    
  4. Push your changes back upstream, replacing filetype with the name of your branch:

    git push origin filetype
    

Stop tracking a file with Git LFS

When you stop tracking a file with Git LFS, the file remains on disk because it remains part of your repository’s history. To understand why, see Delete a Git LFS file from repository history.

Prerequisites:

  • You have downloaded and installed the appropriate version of the CLI extension for Git LFS for your operating system.
  • You have installed the Git LFS pre-push hook by running git lfs install in the root directory of your repository.

To stop tracking a file with Git LFS:

  1. Run the git lfs untrack command and provide the path to the file:

    git lfs untrack doc/example.iso
    
  2. Use the touch command to convert it back to a standard file:

    touch doc/example.iso
    
  3. Tell Git to track the changes to the file:

    git add .
    
  4. Commit and push your changes.
  5. Create a merge request and request a review.
  6. After you get the required approvals, merge the request into the target branch.

If you delete an object (example.iso) tracked by Git LFS, but don’t use the git lfs untrack command, example.iso shows as modified in git status.

Stop tracking all files of a single type

Prerequisites:

  • You have downloaded and installed the appropriate version of the CLI extension for Git LFS for your operating system.
  • You have installed the Git LFS pre-push hook by running git lfs install in the root directory of your repository.

To stop tracking all files of a particular type in Git LFS:

  1. Run the git lfs untrack command and provide the file type to stop tracking:

    git lfs untrack "*.iso"
    
  2. Use the touch command to convert the files back to standard files:

    touch *.iso
    
  3. Tell Git to track the changes to the files:

    git add .
    
  4. Commit and push your changes.
  5. Create a merge request and request a review.
  6. After you get the required approvals, merge the request into the target branch.

Enable or disable Git LFS for a project

Git LFS is enabled by default for both self-managed instances and GitLab.com.

Prerequisites:

  • You must have at least the Developer role for the project.

To enable or disable Git LFS at the project level:

  1. On the left sidebar, select Search or go to and find your project.
  2. Select Settings > General.
  3. Expand the Visibility, project features, permissions section.
  4. Select the Git Large File Storage (LFS) toggle.
  5. Select Save changes.

Clone a repository that uses Git LFS

When you clone a repository that uses Git LFS, Git detects the LFS-tracked files and clones them over HTTPS. If you run git clone with a SSH URL, like user@hostname.com:group/project.git, you must enter your GitLab credentials again for HTTPS authentication.

By default, Git LFS operations occur over HTTPS, even when Git communicates with your repository over SSH. In GitLab 17.2, pure SSH support for LFS was introduced. For information on how to enable this feature, see Pure SSH transfer protocol.

To fetch new LFS objects for a repository you have already cloned, run this command:

git lfs fetch origin main

Migrate an existing repository to Git LFS

Read the git-lfs-migrate documentation on how to migrate an existing Git repository with Git LFS.

Delete a Git LFS file from repository history

It’s important to understand the differences between untracking a file in Git LFS and deleting a file:

  • Untrack: The file remains on disk and in your repository history. If users check out historical branches or tags, they still need the LFS version of the file.
  • Delete: The file is removed but remains in your repository history.

To delete a tracked file with Git LFS, see Remove a file.

To completely expunge all history of a file, past and present, see Delete sensitive information from commits.

caution
Expunging file history requires rewriting Git history. This action is destructive and irreversible.

Reduce repository size after removing large files

If you need to remove large files from your repository’s history, to reduce the total size of your repository, see Reduce repository size.