Skip to main content

Client to control a remote-docker agent

Project description

Remote Docker on AWS for local development

PyPI version Python versions

Use docker to develop services, but without the overhead of running docker on your machine! This is a development tool that you should use if your machine is low performance, or if you are running many docker services.

Why is this useful?

Frees up your local machine for useful tasks such as running your code editor, browser, and email, leaving running Docker to a dedicated server instance. The result is that your local machine functions faster, uses up less disk space, and consumes less power. MacOS users will also see noticeable speed improvements since Docker on Linux (which is what the remote hosts runs) is much more performant.

The downsides:

  • SSH tunnel communication is slower than local communication. However using an AWS region with low ping makes the latency unnoticeable. Find the region fastest for you using this site
  • Some more setup required to get everything configured properly and running (tunneling ports, syncing file changes)
  • Running the ec2 instance incurs an additional cost over running locally, although a t3.medium instance in Canada only costs just under 5 cents/hour

How it works: two processes are run, a sync and a tunnel process.

  • The sync process keeps local and remote files in sync so that the docker process run remotely can use docker volumes transparently
  • The tunnel process forwards ports needed so your local system can communicate with docker, plus additional ports as required, such as port 443 for browser communication

Setup

  1. First login to your AWS account and create access keys to access AWS through the CLI

    You will need the following IAM policies:

    • AmazonEC2FullAccess
    • AWSCloudFormationFullAccess

    And now in your terminal:

    # Replace josh with your name
    # You will need to setup an AWS account if you don't have one
    # and create access key credentials
    
    aws configure --profile josh
    export AWS_PROFILE=josh
    
  2. Install pre-requisites

    Have Homebrew (Available on both macOS and Linux now!)

    Have pipx

    pipx install remote-docker-aws
    pipx install unison-gitignore
    
    # Install unison sync utility
    brew install unison
    
    # Install file-watcher driver for unison
    # On MacOS:
    brew install autozimu/homebrew-formulas/unison-fsmonitor
    
    # Or, on Linux since the above formula doesn't work:
    brew install eugenmayer/dockersync/unox
    
  3. Generate and upload a keypair to AWS

    # Note: bash users can use `rd` instead of `remote-docker-aws`. zsh users cannot since zsh aliases `rd` to `rmdir` (!)
    remote-docker-aws create-keypair
    
  4. Create the ec2 instance

    remote-docker-aws create
    

Daily Running

  1. Start the remote-docker ec2 instance

    remote-docker-aws start
    

    This will automatically switch the docker context for you. If you want to switch back to the default agent run docker context use default

  2. In one terminal start the tunnel so that the ports you need to connect to are exposed

    remote-docker-aws tunnel
    
    # Usually it's preferable just to forward the ports to same port
    # so eg. with mysql on docker exposing port 3306 and nginx on docker exposing port 80:
    remote-docker-aws tunnel -l 80:80 -l 3306:3306
    
    # You can forward remote ports as needed with the "-r" option:
    # which can be used so the docker instance can access services running locally (eg. webpack)
    remote-docker-aws tunnel -r 8080:8080
    
  3. In another terminal sync file changes to the remote instance:

    # Add any more paths you need to sync here, or add them to the config file
    # You will need to sync directories that are mounted as volumes by docker
    remote-docker-aws sync ~/blog
    
    # If watched directories are supplied in ~/.remote-docker.config.json
    # then simply call:
    remote-docker-aws sync
    
  4. Develop and code! All services should be accessible and usable as usual (eg: docker ps, docker-compose up, etc.) as long as you are running remote-docker-aws tunnel and are forwarding the ports you need

  5. When you're done for the day don't forget to stop the instance to save money:

    remote-docker-aws stop
    

Config File

Looks for a config file at the path ~/.remote-docker.config.json by default, which can be overriden by passing --config-path. The config file is not necessary and CLI usage is possible without it as long as AWS_PROFILE and AWS_REGION environment variables are set

An example .remote-docker.config.json file:

{
    "key_path": "~/.ssh/id_rsa_remote_docker",
    "sync_ignore_patterns_git": [
        "**/*.idea/",
        "**/*.git/",
        "**/*~",
        "**/*.sw[pon]"
    ],
    "profiles": {
        "blog": {
            "sync_ignore_patterns_git": [
                "**/notes/"
            ],
            "remote_port_forwards": {
                "local-webpack-app": {"8080": "8080"}
            },
            "local_port_forwards": {
                "blog_app": {"443": "443", "80":  "8000"},
                "blog_db": {"3306": "3306"}
            },
            "watched_directories": [
                "~/.aws",
                "~/blog"
            ]
        }
    },
    "default_profile": "blog"
}
Usage: remote-docker-aws [OPTIONS] COMMAND [ARGS]...

Options:
  --profile TEXT      Name of the remote-docker-aws profile to use
  --config-path TEXT  Path of the remote-docker-aws JSON config

The current configurable values are:

aws_region (takes precedence over AWS_REGION and .aws/config)

  • The region to create the instance in

instance_type

  • Type of ec2 instance, defaults to: t3.medium

key_path

  • defaults to: ~/.ssh/id_rsa_remote_docker

local_port_forwards

  • defaults to: {}
  • Object containing label -> port mapping objects for opening the ports on the remote host. A mapping of "local_port_forwards": {"my_app": {"80": "8080"}} will open port 80 on your local machine and point it to port 8080 of the remote-docker instance (which ostensibly a container is listening on). The name doesn't do anything except help legibility.

remote_port_forwards

  • defaults to: {}

  • Similar to local_port_forwards except will open the port on the remote instance.

    This is useful to have frontend webpack apps accessible on the remote host

sync_ignore_patterns_git

  • defaults to: []
  • use .gitignore syntax, and make sure to use the directory wildcard as needed

user_id

  • defaults to None
  • Used to uniquely identify the instance, this is useful if multiple remote-docker agents will be created in the same AWS account

watched_directories

  • defaults to: []
  • list of paths to watch by remote-docker-aws sync

volume_size

  • defaults to: 30 (GB)
  • Size of the ec2 volume.

Profiles are a way to organize and override settings for different projects. Values nested in a profile override the values defined outside a profile, except for lists and dictionaries which are merged with the values outside the profile

Cost

A t3.medium instance on ca-central-1 currently costs $0.046 /hour. See current prices

Nothing else used should incur any cost with reasonable usage

Notes

  • See remote-docker-aws --help for more information on the commands available
  • The unison version running on the server and running locally have to match. If one of them updates to a newer version, you should update the other. This can be done locally via brew upgrade unison, and to update the remote unison version: rd ssh then brew upgrade unison

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

remote-docker-aws-3.0.1.tar.gz (25.2 kB view details)

Uploaded Source

Built Distribution

remote_docker_aws-3.0.1-py3-none-any.whl (18.6 kB view details)

Uploaded Python 3

File details

Details for the file remote-docker-aws-3.0.1.tar.gz.

File metadata

  • Download URL: remote-docker-aws-3.0.1.tar.gz
  • Upload date:
  • Size: 25.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.11.1

File hashes

Hashes for remote-docker-aws-3.0.1.tar.gz
Algorithm Hash digest
SHA256 64469d58548aaf90091380efde55b2e7808ad4627525f26e8e0b4320575f3062
MD5 d45f281f0ee318e4cf0cfddea7d369fe
BLAKE2b-256 2736667909622eb3fd3bc9180db2226bc13b630ce86a60a30f2f524e9bd5503a

See more details on using hashes here.

File details

Details for the file remote_docker_aws-3.0.1-py3-none-any.whl.

File metadata

File hashes

Hashes for remote_docker_aws-3.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 aa1638a527776b853c3f2dfd55a643e2266fd99e9206369c5560b859933be232
MD5 350e805e195e941523c1800f138a3d00
BLAKE2b-256 9fb7d66f0e4e7237f13b4b3350ac0313557020edaa00777d5972207a55315bbd

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page