Simple non-containerize application with GitHub actions

Open Table of contents

Intro

Intro

I worked on many projects with Github as DevOps in my team. I have a responsibility to deliver applications from development to production. At first, I did it manually. I started to feel this is always reprising every time there is a change in the application.

thumb

Our production environment still uses a bare metal server. The application ran on VMs created in proxmox. Production ran a simple stack. but if a change or feature is added to the Github repository my dev friends say, “Hey I added new feature, can you deploy it?” and it keeps repeating. until I found solution to create Github actions that can continuously check and deliver the application.

provisioning DB

Our application depends on PostgreSQL. I used containerize DB because it’s simple to provisioning and you know.. I want to keep the server clean. to realize it creates a docker-compose file to automate provisioning.

	version: '3.8'

	services:

	db:
	container_name: postgres_db
	image: postgres:13-alpine
	restart: always
	environment:
	POSTGRES_USER: dbadmin
	POSTGRES_PASSWORD: dbpassword
	POSTGRES_DB: maindb
	TZ: Asia/Jakarta
	volumes:
	- postgres_data:/var/lib/postgresql/data/
	networks:
	vpcbr:
	ipv4_address: 10.10.1.2 # better to keep it static

	volumes:
	postgres_data:

	networks:
	vpcbr:
	driver: bridge
	ipam:
	config:
	- subnet: 10.10.1.0/24
	gateway: 10.10.1.1

view raw docker-compose.db.yml hosted with ❤ by GitHub

let’s make this up in daemon mode.

docker-compose -f docker-compose.db.yml up -d

Create deploy.sh

to automate working my repeating task in deployment, I use simple bash scripting.
this script would be run every deployment.

	#!/bin/bash
	# deploy.sh uncontainerized ci/cd

	# stashing repository
	# make sure branch isn't conflict
	# bring it back, and pull new change
	echo "git stashing repository..."
	git stash
	if [ $? -ne 0 ];
	then
	echo "error when stashing repository..."
	exit 1
	fi

	# pull repository
	echo "git pull..."
	git pull
	if [ $? -ne 0 ];
	then
	echo "error when pulling repository..."
	exit 1
	fi

	# change env-dev to env-prod in app/settings.py
	# make sure your program config is ready to production env
	echo "changing settings file"
	sed -i 's/.env-dev/.env-prod/g' app/settings.py
	sed -i 's/localhost:8000/example.domain.com/g' static/js/vueApp.js
	if [ $? -ne 0 ];
	then
	echo "error when sed the text..."
	exit 1
	fi

	# collectstatic
	source prod/bin/activate

	echo "Install depedencies.."
	pip install -r requirements.txt
	if [ $? -ne 0 ];
	then
	echo "Problem when installing depedencies, check requirements.txt.."
	exit 1
	fi

	python manage.py migrate --no-input
	if [ $? -ne 0 ];
	then
	echo "db connection failed, please check password and user correctly"
	exit 2
	fi

	python manage.py collectstatic --no-input
	if [ $? -ne 0 ];
	then
	echo "staticfile error please check path correctly"
	exit 2
	fi

	# restarting gunicorn and nginx
	sudo gunicornrestart
	if [ $? -ne 0 ];
	then
	echo "gunicorn restart failed please check it out!"
	fi

	sudo nginxrestart
	if [ $? -ne 0 ];
	then
	echo "nginx restart failed please check it out!"
	fi

	# check server status
	CODE=$(curl -s -w "%{http_code}\n" https://example.domain.com/ -o /dev/null)
	GIT_SHA=$(git rev-parse HEAD \| cut -c 1-7)
	NODE=$(hostname)
	NGINX=$(systemctl status nginx \| awk '/Active/ {print $2" "$3}')
	GUNI=$(systemctl status gunicorn \| awk '/Active/ {print $2" "$3}')

	echo "========================================="
	echo "deploy success!"
	echo "on node -> $NODE"
	echo "on commit -> $GIT_SHA"
	echo "curl status -> $CODE"
	echo "nginx status -> $NGINX"
	echo "gunicorn status -> $GUNI"
	echo "========================================="

view raw deploy.sh hosted with ❤ by GitHub

the working directory would be stashing before pulling the repository, this is done so that there is no conflict merge. and then did pulling repo. in production have little change configuration settings like DB. I use sed to change the settings file with one line. make sure already migrated. it always restarts service gunicorn and nginx to minimalization failure. see how to create gunicorn.service file.

Setup Github actions workflow

The first thing we should do is create GitHub secrets. Github secrets are encrypted secrets allowing us to store sensitive information in your organization, repository, or repository environments. learn more here.

SSH_KEY         -> ssh private key (for server connection)
SSH_PORT        -> ssh port
SSH_PASSPHRASE  -> optional but highly recommended
SSH_HOST        -> host ssh server (target server)

may you ask “Why Host and Port should be in github secrets?” i recommend to keep all data of server is encyrpted and secret for security purpose. next, what I did is set up Github actions to trigger change and run deploy.sh over ssh. here are my Github actions

	name: Deployment

	on:
	workflow_run:
	workflows:
	- "Django CI"
	types:
	- completed

	jobs:
	build:
	name: deploy to server over ssh command
	runs-on: ubuntu-20.04

	steps:
	- name: Deploy to Production
	uses: appleboy/ssh-action@v0.1.4
	with:
	host: ${{ secrets.SSH_PROD_HOST }}
	username: ${{ secrets.SSH_PROD_USER }}
	port: ${{ secrets.SSH_PROD_PORT }}
	key: ${{ secrets.SSH_PROD_KEY }}
	passphrase: ${{ secrets.SSH_PROD_PASSPHRASE }}
	script: \|
	cd path/to/work
	bash deploy.sh

view raw deploy-django.yml hosted with ❤ by GitHub