AWS DevOps

https://wilsonmar.github.io/aws-devops/

Developer Operations for you

Overview

CodeCommit, CodePipeline, Code Deploy
AMIs
More on Amazon
More resources
More on DevOps

This tutorial focuses on continuous integration and deployment.

Amazon defines “DevOps is the combination of cultural philosophies, practices, and tools that increases an organization’s ability to deliver applications and services at high velocity.”

CodeCommit, CodePipeline, Code Deploy

aws devops 201606-650x209-i11

Setup instances

In IAM Service, create Role “codedeploy”.
Create CDInstanceRole
In Compute EC2 service, launch Amazon Linux, t2.micro, 2 instances, using the role created above. In Advanced Details, paste script from https://gist.github.com/mikepfeiffer/4d9386afdcceaf29493a

EC2 UserData script to install CodeDeploy agent:
```
#!/bin/bash
yum install -y aws-cli
cd /home/ec2-user/
aws s3 cp 's3://aws-codedeploy-us-east-1/latest/codedeploy-agent.noarch.rpm' . --region us-east-1
yum -y install codedeploy-agent.noarch.rpm
```
CUSTOMIZE folder, region mentionedtwice.
Tag instances with name “Dev” for Development.
Add a Security Group Role for HTTP. No SSH.

AWS CodeDeploy Setup

Got to AWS CodeDeploy service, Get Started Now.
Custom deployment.
Specify an Application Name and Deployment Group Name according to your organization’s naming standards.
Select Tag Type “Amazon EC2” value “Dev” specified for 2 instances in a step above.
Deployment Config - AllAtOnce (instead of Half at a time).
No Triggers.
Select a Service Role ARN defined in a prior step. Create Application.

The Console provides code to deploy from a S3 bucket.

AWS CodePipeline

Go to AWS CodePipeline service, Get Started.
Specify a Pipeline name according to your organization’s naming standards. (“Pipeline1”)
Select Source Provider: GitHub (NOT Amazon S3). Click Connect to GitHub.
Select a Repository and Branch from the GitHub account authenticated.
Select Deployment provider AWS CodeDeploy (NOT AWS Elastic Beanstalk).
Supply AWS CodeDeploy Application Name and Deployment group from earlier.
Do not define Build Stage (until we have a build).
Create Service Role using default name “AWS-CodePipeline-Service”. View Policy Document to review Actions allowed the role:
Review Pipeline summary.

View app deployed

In EC2, get the Public DNS address (such as “ec2-11-222-177-132-us-west-2-compute.amazonaws.com”).
Paste URL in an internet browser.

It should respond with “Congratualations”.

Make Change

Commit.
Detect a change.
View app deployed again.
Create a Deployment Group or Autoscaling Group
CodePipeline

appspec.yml file in the root folder in source code repo

version: 0.0
os: linux
files:
         - source: /index.html
 destination: /var/www/html/
hooks:
  BeforeInstall:
 - location: scripts/install_dependencies
   timeout: 300
   runas: root
 - location: scripts/start_server
   timeout: 300
   runas: root
  ApplicationStop:
 - location: scripts/stop_server
   timeout: 300
   runas: root

For sample application, it’s just a single index.html file containing CSS, no JavaScript.

https://github.com/mikepfeiffer/aws-codedeploy-linux/blob/master/index.html
Install dependencies
```
#!/bin/bash
yum install -y httpd
```
Start server
```
#!/bin/bash
service httpd start
```

Stop server:

#!/bin/bash
isExistApp = `pgrep httpd`
if [[ -n  $isExistApp ]]; then
    service httpd stop        
fi

using AWS Code Services

CodeDeploy agent in EC2 Deploy Group

Amazon Route 53 globomantics.com
https://github.com/mikepfeiffer/PowerShell

AMIs

Netflix/aminator open-sourced their Python tool for creating EBS-backed AMIs. It’s described in this blog from 2013.

“We knew that application startup latency would be very important, especially during scale-up operations.”

Building a server from installers in S3 can be time-consuming because it take so much I/O.

To save time, their strategy is to create a Base AMI by taking a snapshot of the root volume and make it available as an EBS volume that can be used to launch an EC2 instance.

This is simpler than Packer from Hashicorp.

More on Amazon

This is one of a series on Amazon:

More resources

Justin Manga (@jmenga, pseudo.co.de) in video course Continuous Delivery using Docker and Ansible shows learners how to create a Python app, then Jenkins 2 Pipeline plugin in a container using Cloud Formation files. deploy to a EC2 Container Services holding Debian instances.

Since ECS has Group of 497:

ARG DOCKER_GID=497
RUN groupadd -g ${DOCKER_GID:-497} docker
ARG DOCKER_ENGINE=1.10.2
ARG DOCKER_COMPOSE=1.6.2

http://blog.serverbooter.com/blog/2013/10/24/simple-clouformation-with-multiple-aws-accounts/

More on DevOps

This is one of a series on DevOps:

Wilson Mar