crusher

README

crusher

Configuration Management Tool

Intro

crusher is a minimalist configuration-management tool. It aims to provide much of the flexibility of the available off-the-shelf options, with as little overhead as possible.

Crusher breaks down the configuration process into 5 basic pieces:

1. Pre-configuration Commands - commands you need to run before anything else (adding package repo's creating folders, etc.)
2. Apt-Get Packages - installing necessary software packages
3. Configuration Files - (/etc/ files)
4. Content - (websites, data files)
5. Post-configuration Commands - commands you need to run after everything else (restating services, reloading configuration files, etc.)

This project started out as a code challenge as part of a job interview for Slack, but I liked it so much that I've kept it going.

Requirements

To use this tool, you must have the ability to ssh to a host, have permission to write to the servers temp directory, and run commands with elevated privileges.

Use Cases

crusher can be used as both a centralized and distributed tool for setting up new servers.

• Centralized: crusher manages a list of remote servers that it saves in your users home directory (~/.crusher). The remote-configure command targets remote servers based on name or spec, and runs configuration tasks on all of them asynchronously.

• Distributed: compile crusher and put it at the base of a git repo containing your spec files. New servers can be launched with a script to check out your repo and run crushers local-configure command to configure themselves.

Installation

curl -s http://dl.sudoba.sh/get/crusher | sh

Commands

The available commands can be printed by running $ crusher or $ crusher --help:

$ crusher --help
crusher - 1.0

Usage:
   crusher [global options] command [command options] [arguments...]

Commands:
   list-servers, l			List all configured remote servers
   remote-configure, rc		Configure one or many remote servers
   local-configure, lc		Configure this local machine with a given spec
   add-server, a			Add a new remote server to the config
   delete-server, d			Delete a remote server from the config
   available-specs, s		List all available specs
   show-spec, ss			Show what a given spec will build
   help, h					Shows a list of commands or help for one command

Global Options:
   --version, -v			print the version
   --help, -h				show help

Help for specific commands can be printed by passing the --help flag to the command

$ crusher remote-configure --help
Usage:
   remote-configure search [--flags]

Arguments:
   search					The server or spec group to remote configure

Flags:


Example:
   crusher remote-configure hello_world

What's a Spec?

A .spec file (short for specification), along with its config and content folders, contain the building blocks of a server configuration. Specs contain a list of packages to install, configuration and content files along with their destinations, and commands to run during the configuration job.

An example of a spec that installs php7:

NAME = php

VERSION = 1
REQUIRES =

[PACKAGES]
	apt_get = php7.0-fpm, php7.0-cli, php7.0-curl, php7.0-gd, php7.0-intl php7.0-mysql, php-memcache, php7.0-xml, php7.0-mbstring, php7.0-mcrypt, php7.0-xmlrpc

[CONFIGS]
	debian_root = "/etc/"
	skip_interpolate = true

[COMMANDS]
	pre = "sudo chmod -R 775 ./specs/php/scripts/, sudo add-apt-repository -y ppa:ondrej/php"
	post = "sudo service php7.0-fpm restart"

Specs can require other specs, to link smaller building blocks into more complex configurations. Check out hello_word.spec in the example-specs folder for a simple example.

By default, crusher will look for Specs in the following directories, in order, overwriting previously found specs with the same name:

1. $GOPATH/src/github.com/murdinc/crusher/example-specs/
2. /etc/crusher/specs/
3. ~/crusher/specs/
4. ./specs/

Running a Spec against a server looks a little like this:

1. Install all required packages from the spec and all of its required specs
2. Transfer/Copy all of the config and content files of the spec and its required specs
3. Run post-config commands of the spec and its required specs

Hello World Example:

This example spec installs nginx and php5-fpm, and serves "Hello World!" from port 80.

1. Run the remote-configure command, passing in hello_world as the search option:

$ crusher remote-configure hello_world

1. crusher knows you haven't configured any remote servers yet, and asks you to set some up first:

   

2. Run the same command again, and this time it will find the servers you have set up, and run the spec configurations against them:

   

3. Your servers should now be serving "Hello World!" from port 80.

Crusher?

This was a code challenge, and I for some reason immediately thought of the scenario of Wesley Crusher adding a set of commands to the ships computers to automate the set-up new warp engines. Also I needed to call it something.

Tests

There is a very basic test for each of the sub-packages, I hope to expand those after moving the Jobs code into its own sub-package. Running $ go test github.com/murdinc/crusher/... will run all tests for this project.

$ go test github.com/murdinc/crusher/...
ok  	github.com/murdinc/crusher	0.010s
ok  	github.com/murdinc/crusher/config	0.011s
ok  	github.com/murdinc/crusher/servers	0.011s
ok  	github.com/murdinc/crusher/specr	0.013s

Roadmap / Not yet implemented

• SSH Key Authentication (still needs callback func)
• Support for all flavors of servers (not just Ubuntu)
• Finer control over tasks run / incremental changes
• Check and rollback of config changes
• Pull Jobs into its own package as an interface to help declutter the other controllers
• More Tests!
• Lots of sanity checking still needed
• Tab completion