regluit/docs/vagrant_ansible.md

98 lines
4.0 KiB
Markdown
Raw Normal View History

**Instructions on how to run open_source branch**
To use vagrant/ansible to build the `{please|just|}.unglue.it`, you
need to have the following installed:
* [Vagrant](https://www.vagrantup.com/docs/installation/) (e.g., to download: https://www.vagrantup.com/downloads.html) (at least version 1.8.1)
* [Installation — Ansible Documentation](http://docs.ansible.com/ansible/intro_installation.html#latest-releases-via-pip) (version 2+) -- use `pip install ansible`
We also need the `vagrant-aws vagrant plugin:
```
vagrant plugin install vagrant-aws --plugin-version 0.5.0
```
Optionally you can [VirtualBox Oracle VM VirtualBox](https://www.virtualbox.org/wiki/VirtualBox) to enable the build of machines locally.
Layout of important files:
* [Vagrantfile](https://github.com/Gluejar/regluit/blob/1ac55c4f0a6b6a3dfc97652aa5ce33638a6140a1/vagrant/Vagrantfile), which is what `vagrant` looks for and defines various hosts: `please`, `just`, `just2`, `prod`, and `prod2`.
* [dev.yml](https://github.com/Gluejar/regluit/blob/1ac55c4f0a6b6a3dfc97652aa5ce33638a6140a1/vagrant/dev.yml) -- the main ansible playbook that builds the various machines
* `please` is for buiding `please.unglue.it` -- it is a transient machine
* the reason I have a `just` *group* with `just` and `just2` hosts is while one is in production, I build the new one. Once the new one is working, I can `vagrant stop` and then ultimately `vagrant destroy` the old one.
* similar logic for the production *group*. (Note that before I retire a production server, I copy over the logs to S3: [backing up production logs to S3](https://www.evernote.com/shard/s1/sh/f12406a7-de95-4d54-809d-9f3abe8eaabd/f935e813d8f16f25))
You also need AWS keys in the environment. I have my key/secret pair configured with a shell script that I can run -- I've stored this file in `/Volumes/ryvault1/gluejar/other_keys/aws.sh`, stored in an encrypted volume on my laptop. For convenience I link to the file from `~/bin/gj_aws.sh`:
```
#!/bin/bash
export BOTO_CONFIG=/Volumes/ryvault1/gluejar/other_keys/boto.cfg
# rdhyee key: https://console.aws.amazon.com/iam/home?region=us-east-1#/users/rdhyee
# eric: you can use the credentials from https://console.aws.amazon.com/iam/home?region=us-east-1#/users/eric
# note: these credentials are tied to the Gluejar account
export AWS_ACCESS_KEY=[FILL IN]
export AWS_SECRET_KEY=[FILL IN]
# simple_public_vpc
export AWS_VPC_ID=vpc-6f7db10b
# EC2 API tools
export EC2_ACCESS_KEY=$AWS_ACCESS_KEY
export EC2_SECRET_KEY=$AWS_SECRET_KEY
# s3cmd
ln -fs /Volumes/ryvault1/gluejar/s3/s3cfg ~/.s3cfg
export AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY
export AWS_SECRET_ACCESS_KEY=$AWS_SECRET_KEY
# AWS CLI http://docs.aws.amazon.com/cli/latest/userguide/cli-chap-getting-started.html
export AWS_DEFAULT_REGION=us-east-1
export AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY
export AWS_SECRET_ACCESS_KEY=$AWS_SECRET_KEY
# http://docs.aws.amazon.com/cli/latest/userguide/controlling-output.html
export AWS_DEFAULT_OUTPUT="json"
```
e.g.,
```
hyptyposis-2014:vagrant raymondyee$ ls -lt ~/bin/gj_aws.sh
lrwxr-xr-x 1 raymondyee 501 43 Aug 18 2014 /Users/raymondyee/bin/gj_aws.sh -> /Volumes/ryvault1/gluejar/other_keys/aws.sh
```
In the `regluit/vagrant` directory, after I run `~/bin/gj_aws.sh` and `vagrant status`, I get something like (the actual status of various machines can vary):
```
please not created (virtualbox)
just running (aws)
just2 not created (virtualbox)
prod not created (virtualbox)
prod2 running (aws)
localvm not created (virtualbox)
```
Once you have `vagrant status` works, a good place to start is how to build `please` with
```
vagrant up please --provider=aws
```
**For the moment, please leave building just and production to me.**
## running regluit on localhost
In the `vagrant` directory, you can run
```
ansible-playbook create_commonpy.yml
```
to generate `settings/common.py`. You should then be able to proceed as normal.