Opsworks

OpsWorks #

• stack = cookbooks
• layer = frontend, backend, api, rds (how to config to instance)
• app = source code. can deploy and redeploy

Stack #

• I am exploring this cookbook from aws
• The nodejs demo cookbook repo does not include its dependencies
• In the opsworks demo, it uses opsworks-linux-demo-cookbooks-nodejs.tar.gz as the cookbook source
• We need to convert the repo to the archive before using it
• Aftering changing the cookbook source, we need to fetch the cache by running commands on instances

Inside the opsworks-linux-demo-cookbooks-nodejs.tar.gz, we have

nodejs_demo
 recipes
   default.rb

To run this cookbook, we set nodejs_demo::default in the layer

Docker Support #

• eb -> multi docker / docker + elb
• opsworks -> ecs with linux + ec2

Layer #

• auto-healing = restart the instance when losing connection with opsworks agent

• elb

• sg

• ebs

• eip

• cloudwatch

• instance

• rds service layer

• elb service layer

• ecs cluster layer

App #

• source code
• env
• domain
• ssl
• db

source code are placed at /svr/<app_name>

Instance Types #

• 24/7 = normal instance
• time = start at specific period
• load-based = asg

Behaviors on auto-healing between these two instance type

• ebs-backed = re-create the instance
• instance = stop and start

Issue #

Instances are stuck in booting #

Reason: incorrect instance iam role

We can find the error message at

cat tail -f /var/logs/aws/opsworks/*

My instance iam role is incorrect because it should trust ec2.amazonaws.com instead of opsworks.

and I don’t realise that opsworks need 2 iam roles:

• service role: use by opsworks. need to trust opsworks.amazonaws.com and passrole to ec2.amazonaws.com
• instance iam role: use by ec2. need to trust ec2.amazonaws.com and passrole to opsworks.amazonaws.com