Rudolfs

A high-performance, caching Git LFS server with an AWS S3 back-end.

Features

• Multiple backends:

  1. AWS S3 backend with an optional local disk cache.
  2. Local disk backend.

• A configurable local disk cache to speed up downloads (and reduce your S3 bill).

• Corruption-resilient local disk cache. Even if the disk is getting blasted by cosmic rays, it'll find corrupted LFS objects and purge them from the cache transparently. The client should never notice this happening.

• Encryption of LFS objects in both the cache and in permanent storage.

• Separation of GitHub organizations and projects. Just specify the org and project names in the URL and they are automatically created. If two projects share many LFS objects, have them use the same URL to save on storage space.

• A tiny (<10MB) Docker image (jasonwhite0/rudolfs).

The back-end storage code is very modular and composable. PRs for implementing other storage back-ends are welcome. If you begin working on this, please let us know by submitting an issue.

Non-Features

• There is no client authentication. This is meant to be run in an internal network with clients you trust, not on the internet with malicious actors.

Running It

Generate an encryption key (optional)

If configured, all LFS objects are encrypted with the xchacha20 symmetric stream cipher. You must generate a 32-byte encryption key before starting the server.

Generating a random key is easy:

openssl rand -hex 32

Keep this secret and save it in a password manager so you don't lose it. We will pass this to the server below via the --key option. If the --key option is not specified, then the LFS objects are not encrypted.

Note:

• If the key ever changes (or if encryption is disabled), all existing LFS objects will become garbage. When the Git LFS client attempts to download them, the SHA256 verification step will fail.
• Likewise, if encryption is later enabled after it has been disabled, all existing unencrypted LFS objects will be seen as garbage.
• LFS objects in both the cache and in permanent storage are encrypted. However, objects are decrypted before being sent to the LFS client, so take any necessary precautions to keep your intellectual property safe.

Development

For testing during development, it is easiest to run it with Cargo. Create a file called test.sh (this path is already ignored by .gitignore):

# Your AWS credentials.
export AWS_ACCESS_KEY_ID=XXXXXXXXXXXXXXXXXXXX
export AWS_SECRET_ACCESS_KEY=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
export AWS_DEFAULT_REGION=us-west-1

# Change this to the output of `openssl rand -hex 32`.
KEY=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

cargo run -- \
  --cache-dir cache \
  --host localhost:8080 \
  --max-cache-size 10GiB \
  --key $KEY \
  s3 \
  --bucket foobar

If you just need to use the local disk as the backend, use the following bash.

# Change this to the output of `openssl rand -hex 32`.
export RUDOLFS_KEY=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

cargo run -- --port 8080 local --path=/data

Note: Always use a different S3 bucket, cache directory, and encryption key than what you use in your production environment.

Warning: This server may not be accessible from other machines. Specifying --host localhost:8080 will often bind the server to an internal-only loopback network interface (i.e., if localhost resolves to 127.0.0.1 or [::1]). Thus, to make the server accessible from the outside world, specify --host 0.0.0.0:8080 or just --port 8080 (the default IP the server will bind to is 0.0.0.0). IP 0.0.0.0 means the server shall try to bind to all available network interfaces, both internal and external. See #38 (comment) for more information.

Production

To run in a production environment, it is easiest to use docker-compose:

1. Create a .env file next to docker-compose.yml with the configuration variables:

   AWS_ACCESS_KEY_ID=XXXXXXXXXXXXXXXXXXXQ
   AWS_SECRET_ACCESS_KEY=XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
   AWS_DEFAULT_REGION=us-west-1
   LFS_ENCRYPTION_KEY=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
   LFS_S3_BUCKET=my-bucket
   LFS_MAX_CACHE_SIZE=10GB
   

2. Use the provided docker-compose.yml file to run a production environment:

   docker-compose up -d
   
   # use minio yml
   docker-compose -f ./docker-compose.minio.yml up -d
   
   # use local disk yml
   docker-compose -f ./docker-compose.local.yml up -d

3. [Optional]: It is best to use nginx as a reverse proxy for this server. Use it to enable TLS. How to configure this is better covered by other tutorials on the internet.

Note:

• A bigger cache is (almost) always better. Try to use ~85% of the available disk space.
• The cache data is stored in a Docker volume named rudolfs_data. If you want to delete it, run docker volume rm rudolfs_data.

AWS Credentials

AWS credentials must be provided to the server so that it can make requests to the S3 bucket specified on the command line (with --bucket).

Your AWS credentials will be searched for in the following order:

1. Environment variables: AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY
2. AWS credentials file. Usually located at ~/.aws/credentials.
3. IAM instance profile. Will only work if running on an EC2 instance with an instance profile/role.

The AWS region is read from the AWS_DEFAULT_REGION or AWS_REGION environment variable. If it is malformed, it will fall back to us-east-1. If it is not present it will fall back on the value associated with the current profile in ~/.aws/config or the file specified by the AWS_CONFIG_FILE environment variable. If that is malformed or absent it will fall back to us-east-1.

Client Configuration

Add a file named .lfsconfig to the root of your Git repository and commit it so everyone is using the same LFS server:

[lfs]
url = "http://gitlfs.example.com:8080/api/my-org/my-project"
              ─────────┬──────── ──┬─ ─┬─ ───┬── ─────┬────
                       │           │   │     │        └ Replace with your project's name
                       │           │   │     └ Replace with your organization name   
                       │           │   └ Required to be "api"
                       │           └ The port your server started with
                       └ The host name of your server

Optionally, I also recommend changing these global settings to speed things up:

# Increase the number of worker threads
git config --global lfs.concurrenttransfers 64

# Use a global LFS cache to make re-cloning faster
git config --global lfs.storage ~/.cache/lfs

License

MIT License

Thanks

This was developed at Environmental Systems Research Institute (Esri) who have graciously allowed me to retain the copyright and publish it as open source software.