In this article, you will find: Git LFS introduction, principle analysis, download and install, and basic use methods

1. Brief introduction

During the development process, you may need to version large binaries such as images, audio and video, designs, and so on

But any minor change to these large files can result in huge commits, causing the warehouse to expand rapidly and ultimately make it unusable

Cloning the repository is more time consuming because the entire history of the repository needs to be transferred to the client during the cloning process

Fortunately, a Git extension (Git Large File Storage, Git LFS) developed by Github solves this problem nicely


2. Principle analysis

How does Git LFS solve the problem?

Git LFS doesn’t magically handle large files, it just shifts the burden to remote servers, keeping the local repository relatively lean

First, it stores large files in space outside of the main repository, where only lightweight Pointers are kept, which greatly reduces the size of the repository

Second, large files are downloaded at Checkout, not clone or fetch, so you only need to download the required version instead of the historical version


Here’s a more detailed breakdown (see Atlassian Blog, Atlassian is one of the leading developers of Git LFS)

  1. When you add a file to the repository (git addGit LFS replaces its content with Pointers and stores the real content in the local LFS cache located in the repository.git/lfs/objectsDirectory)
  2. When you push a new commit to the remote end (git push), the Git LFS file referenced by the new commit will be transferred directly from the local LFS cache to the remote LFS store bound to the Git repository.
  3. When you check out a commit that contains a Git LFS pointer (git checkout), the pointer file will be replaced with a file in the local LFS cache or downloaded from the remote LFS store


3. Download and install

The following is the download method for Windows users, other system users may be slightly different, please refer to here

  1. Download the installation package (this is v2.13.2, you can also find the latest download on the official website)
  2. Run the installation package and install Git LFS
  3. Execute at terminalgit lfs installInitial Git LFS

Git LFS can be used to manage large files using Git LFS


4. Basic use

Common commands:

  • Git LFS track: adds the specified file mode to Git LFS management
  • Git LFS untrack: Removes the specified file mode from Git LFS management
  • git lfs track: Displays the list of matches currently managed by Git LFS
  • git lfs status: Displays the status of the objects managed by Git LFS
  • Git LFS clone: Clone the remote repository


Create a new Git LFS repository

After you initialize a Git repository with git Init, you also need to run Git LFS install

This installs a pre-push hook in the repository that will transfer the LFS file to the server when git push is performed


Clone an existing Git LFS repository

You can still clone a repository using git clone, and Git will check out the default branches and download all the associated LFS files one by one

A better approach, however, would be to use the git LFS clone command explicitly for better performance

This is because instead of downloading LFS files one at a time, the command downloads them in batches after checkout, which takes full advantage of parallel downloads


(3) Tracking files

You can use git LFS track < file mode > to add the specified file mode to git LFS management

After running the above command, a.gitAttributes file appears in the directory that holds information about the managed content

Git LFS automatically creates and updates the.gitAttributes file, but you’ll need to manually commit the changes to the repository

If you no longer need to manage a file mode, you can use the git LFS untrack < file mode > command to remove it from git LFS management


(4) Check the status

There are two ways to view the status of Git LFS management content

Git LFS track displays the list of matches managed by git LFS

Git LFS status is used to check the status of the objects managed by git LFS


5. Other supplements

Github already forbids uploading large files larger than 100 megabytes. What if you’ve already submitted such files locally?

An effective solution is to use the Git filter-repo tool to weed out all historical commits that do not meet the requirements

Start by installing Git filter-repo via PIP

pip install git-filter-repo
Copy the code

You can then use this tool to weed out files larger than 100M from all historical commits

git filter-repo --strip-blobs-bigger-than 100M
Copy the code

Git filter-repo can be used for more specific purposes in the documentation