altitudelabs / devops Goto Github PK

requiring sudo for the tasks changes numerous environment of the instance

• sourcing env variables via ~/.profile
• install & running node
• etc

currently ec2 module in aws.yml playbook take AWS credential from environment variables, which means that we have to change environment variables every time we work on different projects
and place AWS credentials at #38

vm open port number
godaddy alias (or a record) point to ip address
nginx server block in sites_available for port number to xxx.altitudelabs.com

like

     server {
         listen 80;
         server_name 54.179.128.50;

         return 301 $scheme://altitudelabs.com$request_uri;
     }
     server {
         listen 80;
         server_name www.altitudelabs.com;

         return 301 $scheme://altitudelabs.com$request_uri;
     }

• Abstract variables into env_vars folder for global variables that will be used in roles
• Deploy-node-app role should be able to deploy Ghost blog from specified git repo. Create a bare repo with a post-receive hook such that user can type git push staging master to deploy app

Steps in Ansible:

1. create folder /home/ubuntu/repos/{{project_name}}
2. cd /home/ubuntu/repos/{{project_name}}
3. git --bare init
4. copy post-receive template to /home/ubuntu/repos/{{project_name}}/hooks/post-receive
5. create folder /home/ubuntu/{{ project_name }}

Then to deploy from local computer, do:
git remote add deploy ubuntu@{{ instance_ip }}:/home/ubuntu/repos/{{ project_name }}
git push deploy master --force

• Deploy-blog should be able to deploy Ghost blog from specified git repo using similar method
• Should install gulp, grunt, ruby, sass, bower (see issue #1)

Allow to create new group of instance easily, like

./devops add vms-group-a 5

to create 5 ec2 instances as vms-group-a.

do sync with EC2 with

./devops sync

or/and do sync with EC2 before running reprovision playbook,

1. get list of instances from ec2 with same project name
2. see if there are new groups which is not in devops-host, add new groups
3. see if there are old groups which is not exist in AWS ec2, remove from devops-host
4. Update host list for each group
5. Warn the user if his target host group is not exist anymore

now all playbooks are bounded to devops proj
however devops proj should only provide general playbooks as library
proj itself should be able to contain its own playbooks

node:
    npm:

should be

node:
npm:

Keep the test log result in a file, and print it through ansible

Only for the first time, but we should find a better way to do that

Do sudo ntpdate time.nist.gov,
server time may not be updated, and error may occur.
(for example, AWS SES service does not allow request with wrong timestamp.)

like

---
pm2_start_var:  >-
                  PORT={{ nginx_server_port }}
                  AWS_ACCESS_KEY=xxxxxxx
                  AWS_SECRET_KEY= xxxxxxx
                  BRAINTREE_MERCHANT_ID= xxxxxxx
                  BRAINTREE_PUBLIC_KEY= xxxxxxx
                  BRAINTREE_PRIVATE_KEY= xxxxxxx

Currently we can do that with ./devops deploy --extra-vars="@../devops-ignore.yml"
But better to have a short shell script to do that

now deployment involves all launched hosts
but it should allow deployment to a specific group of hosts, e.g. staging

right now the version control of proj and devops are not sync
should figure out how to make devops into submodule so it wont break the deployment logic whenever devops is updated

npm install -g grunt-cli
npm install -g grunt
npm install -g gulp
npm install -g bower
sudo apt-get install ruby1.9.1 ruby1.9.1-dev
sudo gem install sass