README.md

Running Mountpoint for Amazon S3 in a Docker container

This directory contains two Dockerfiles for building container images to run Mountpoint for Amazon S3. The primary Dockerfile builds an image with the latest official Mountpoint release. The alternative Dockerfile.source builds an image by compiling Mountpoint from the latest source version. We recommend using the primary Dockerfile.

Building the image

First, download the Dockerfile if you don't have it already:

wget https://raw.githubusercontent.com/awslabs/mountpoint-s3/main/docker/Dockerfile

Run this command to build a Docker image:

docker build -t mountpoint-s3 .

To build Mountpoint from source, download Dockerfile.source instead, and pass the -f Dockerfile.source argument to this command.

Running the container

Before running the container, you'll need a way to pass AWS credentials into the container. See AWS credentials in the configuration documentation for more details. We recommend using short-term credentials whenever possible:

• If you run your containers on an EC2 instance, you can associate an IAM role with the instance and Mountpoint for Amazon S3 will automatically assume that role at startup. You may need to increase the hop limit for IMDS requests on your instance for the container to successfully acquire credentials this way.
• Amazon ECS supports short-term credentials by associating an IAM role with your ECS task.

If you need to use long-term credentials, set the AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY environment variables, and pass those variables to Docker with the --env AWS_ACCESS_KEY_ID --env AWS_SECRET_ACCESS_KEY arguments.

Running interactively

To launch Mountpoint in an interactive container, run this command:

docker run -ti --cap-add SYS_ADMIN --device /dev/fuse --entrypoint bash mountpoint-s3

Within the container you can run this command to mount a bucket to the /mnt directory, replacing amzn-s3-demo-bucket with the name of your S3 bucket:

mount-s3 amzn-s3-demo-bucket /mnt

Running as a service

You can also run the Docker container as a service, and access the mounted directory from your host or other Docker containers. To do so, first create a directory /path/to/mount in your host filesystem, and a subdirectory /path/to/mount/bucket to be the target of the mount. Then run this command, replacing amzn-s3-demo-bucket with the name of your S3 bucket, and /path/to/mount with the directory you created:

docker run -d --cap-add SYS_ADMIN --device /dev/fuse \
    --mount type=bind,source=/path/to/mount,target=/mountpoint,bind-propagation=shared \
    mountpoint-s3 \
    amzn-s3-demo-bucket /mountpoint/bucket

Your bucket is now mounted in the /path/to/mount/bucket directory on the host. By default, only the user used by the container (likely root) will have access to the mount. You can make the mount accessible to other users by adding the --allow-other flag to end of the above command. See File and directory permissions in the configuration documentation for more details on managing this permissions issue.

Docker permissions and AWS container services

Mountpoint uses the Linux FUSE subsystem to mount its file system. Running a FUSE file system inside a Docker container requires giving the container elevated root-level privileges to the host instance. The commands above do this with the --cap-add SYS_ADMIN argument to docker run.

These elevated privileges aren't available in AWS-managed container environments like AWS Fargate. They are available in Amazon Elastic Container Service (ECS), but not enabled by default. To enable them, add these fields to the containerDefinitions section of an ECS task definition:

"privileged": true,
"linuxParameters": {
    "devices": [
        {
            "hostPath": "/dev/fuse",
            "containerPath": "/dev/fuse"
        }
    ]
}