Rapid distributed system prototyping

§Getting started

First, install fuge via npm; globally.

npm install -g fuge

§A Micoservice shell?

Yup. Fuge is a shell environment that focuses on a specific system configuration, providing an emulation environment in which services can run in development that is close to a modern container production deployment.


In order to run your system via fuge, you need to create a YAML configuration file. Fuge also integrates with docker-compose. Full documentation on configuring fuge is available here.

Full example demonstration systems are available here: https://github.com/apparatus/fuge-examples.

§Example Configuration

A simple configuration is provided below:

  tail: true
  monitor: true
    - '**/node_modules/**'
    - '**/.git/**'
    - '*.log'
  type: process
  path: ../myservice
  run: 'node index.js'
    - myservice=8000
  type: process
  path: ../webapp
  run: 'npm start'
    - webapp=3000
  image: mongo
  type: container
    - mongo=27017:27017

This configuration defines a front end webapp process, a single microservice and a mongodb container. The system can be started by running:

$ fuge shell <path to config file>
fuge> start all

This will start the mongo container through the local docker API and also the webapp and myservice processes. Fuge will also inject configuration information into these processes as environment variables, tail the logs and watch the file system for changes, restarting the appropriate processes as you write code.

To fully explore the capabilites of Fuge, you are encouraged to run the example systems which can be found here: https://github.com/apparatus/fuge-examples.


Fuge is fully compatible with V1, V2 and V3 docker-compose formats. Fuge services can be specified entirely using docker-compose or entirely in fuge.yml or a mixture of anything in between.

§Command Reference

Fuge can be started by running:

$ fuge shell <path to fuge config file>

The fuge command shell supports the following commands

command action
help display a list of supported commands
ps list status of managed processes and containers
info [process] [full]` show information on a specific process. The full option will output the entire environment that will be injected into this process
start [process] / all start a process or all processes
stop [process] / all stop a process and any associated watcher or all processes
debug [process] start a process in debug mode (only supported for node.js processes)
apply [Shell command] run the specified shell command in the root of each service e.g. apply npm install, apply git status
zone display dns zone information for fuge internal DNS server - requires the dns_enabled setting in fuge configuration file
watch [process] / all turn on watching for a process or for all processes
unwatch [process] / all turn off watching for a process or for all processes
tail [process] / all tail output for a process or for all processes
untail [process] / all end tail output for a specific processes or for all processes
grep 'search string' [process] searches a process’ log or all processes’ logs
exit terminate all managed processes and exit the shell

§Shell pass through

In addition to the above commands, Fuge will pass through all other commands to the underlying shell, for example:

fuge> ps aux | grep -i node

Will bypass the fuge ps command and execute the underlying ps command.

fuge> netstat -an | grep -i listen

Will be passed through to the underlying shell to display a lists of open listening port and so on.


The [apparatus team][] encourage open participation. If you feel you can help in any way, be it with documentation, examples, extra testing, or new features please get in touch.


Copyright the apparatus team 2016, Licensed under [MIT][].

Issues? From spelling errors to broken tutorials and everything in between, report them here.