Yoda

Simple tool to dockerize and manage deployment of your project

What is it?

Yoda helps you to dockerize existing application and automate deployment process.

1. Only BASH. No dependency shit!
2. Requirements: git, docker with compose plugin
3. Its simple like simplicity itself
4. Runs on MACOS and Linux systems

Installation

First you need to install Yoda on your laptop. Its supereasy:

git clone [email protected]:muvon/yoda.git
cd yoda && make check && sudo make install

Remember that you need bash version 4 or higher installed at least and GNU implementation of the sed.

Knowledge requirements

1. Docker and its main concept
2. Image and container understanding
3. You know Docker Compose and its file syntax.

Usage example

OK. You have git repository with your project.
Go into this folder and run this command to initialize environment.

yoda init

Now you will get docker folder created in your project.
Next step is prepare Dockerfile that located in docker/images folder.
You can setup docker build options in file docker/Buildfile.

Now you can add container to your project.

yoda add container-name

Change template for docker-compose.yml file in docker/containers/container-name/container.yml.

We are done. Build it and start with just one command now:

yoda start

Done!

Philosophy

1. You can have several images for single project.
2. Each image you use must have Dockerfile located in docker/images folder and named by convetion: Dockerfile-{name}.
3. You can have several containers depends on one image.
4. Each container has own folder in docker/containers with structure followed by convention in this README.
5. You can setup and use any BASH variables in file docker/env.sh. Its pregenerated for you.
6. Envfile is main file that has all info about what should be built and in which environment, also what server runs which environment for deploy.
7. Each container can be scaled N times and started using single template but different names.
8. You can fully customize deploy, build, compose stages just wrapping in your own script using any language.

Init folder structure

When you do yoda init in your project it creates by default yoda folder. This folder has following structure

Path: containers

When you adding new container folder is created here with same name. For example if you add container with name container. Same folder will appear here.
This folder will contain some files.

What is container.yml

This file contains valid docker compose section for current service. container_name is immutable and declared by Yoda internaly. You can specify image key here with shortcut to image from Buildfile. For example if your Buildfile describe image with key "base" you can put here just image: base and Yoda automatic will replace base to image from build params specified in Buildfile.

Path: env.sh

Here you can declare BASH environment variables and use it everywhere.

For example you can write here IMAGE_NAME to set image name with revision and other staff and use it in Buildfile and container.yml.

If you need custom env.sh file for environment you can just create it with name as env.dev.sh that will extend default env.sh file with new variables defined there.

Take a noticy that environment support namespaces for example production.server1 and production.server2. One server cannot hold 2 namespaces so to use custom env.sh in this case just use env.production.sh without namespace of each server.

Path: Buildfile

Its simple file that have following structure:

base: -t $IMAGE_BASE --compress
db: -t postgres:9.5

Each line contains image name and build args that will be passed in docker build command. See more info in docker build --help.

Path: Envfile

Its simple file YAML like with environment and server description:

user@server: production
production: container1 container2=2
dev: container1

Example file above declare server user@server that will be deployed as production. And production will contain one container1 and two container2 instances.
ANd in dev environment only one container with name container1 will be started.

Or in case of multiservers we use like this

user@server1: production.stack1
user@server2: production.stack2
production.stack1: container1 container2=2
production.stack2: container3 container4
dev: container1

In that case almost the same. Dot (.) allows us to separate namespaces and customize which services we use for one server and which for another. In that case we still have production environment. All that goes after dot (.) is about namespace of environment.

Path: Startfile

This file allow you to manage flow of start your complex service.

dev:
  flow: deploy container2 container1=2
  stop: container2
  wait: deploy
production.ns1:
  flow: deploy container2 container1=2
  stop: container2
  wait: deploy
production.ns2:
  flow: deploy container2 container1=2
  stop: container2
  wait: deploy

• flow - It declares flow of starting services. It contains container or container=chunk where is chunk - number that devide starting by chunks
• stop - each service repsented here will be stopped before start
• wait - After run up command with container sometimes you need to wait for exit code of it. This section declares which containers we should wait for

In this example you yoda start command first will stop container2 if its running.
After that it will start deploy, then container2 and then container1 services will be started as chunks by 2.
Also we wait for exit code for deploy container.

Take a notice in case if you use multiserver and namespaces in environment you have to describe each environment with namespace sepately because all of them independed in flow.

Path: docker/.yodarc

This is locked environment file for yoda inited in current project with yoda version and other useful common files. Plese dont edit this file because it rewrites on yoda upgrade.

If you want you can redefine all variables in env.sh file.

Yoda command line tool usage

yoda command arguments

Commands available:

yoda version

Display current Yoda version

yoda help

Display help information

yoda init [folder]

Prepare dockerized skeleton in project directory

yoda upgrade

Upgrade to new version of initialized Yoda in project.

yoda add [CONTAINER...]

Add container or bunch of containers skeleton to project

yoda delete [CONTAINER...]

Delete container or bunch of existing containers from project

yoda build [options] [IMAGES...]

Build images for current project. You can pass optional images you want to build. Default is every image from Buildfile. Options are:

yoda compose [COMPOSE_SCRIPT]

Display generated docker compose file in stdout.

yoda start [options] [CONTAINER...]

Start all containers or only passed with arguments Options are:

You also can manager flow of start and restart of you containers using Startfile

yoda stop [CONTAINER...]

Stop all containers or only passed with arguments

yoda status

Display current status of services

yoda setup [options]

Setup server before it can be used for environment. You can use only Centos 8 distribution. Otherwise you have to setup it in manual mode. Before deploy put your authorized_keys to server folder.

yoda deploy [options]

Deploy single-node or whole cluster It exit with code 0 for success and 1 for failure (if something went wrong also on one node of all). Options are:

yoda rollback [options]

Rollback to previous revision It exit with code 0 for success and 1 for failure (if something went wrong also on one node of all). Options are the same as for deploy command.

yoda destroy

Remove all created services by start command and all local images with volumes