consul-deployment-agent

A cross platform service that synchronises the set of applications installed on a machine with a manifest stored in a Consul key value store.

See this Overview for more information about what exactly the CDA does.

Package Format

consul-deployment-agent installs applications packaged in the AWS CodeDeploy revision format. This is a .zip file containing your code and an appspec.yml file that describes how to install the application.

Hook Scripts

consul-deployment-agent supports the following subset of the deployment lifecycle hooks:

• BeforeInstall
• AfterInstall
• StartApplication
• ValidateService

Hook Script Execution Environment

consul-deployment-agent sets the following variables in the hook script execution environment

Variable	Value
APPLICATION_ID	?
DEPLOYMENT_BASE_DIR	?
DEPLOYMENT_ID	Environment Manager deployment ID for this application installation
LIFECYCLE_EVENT	The name of the currently executing lifecycle event. One of BeforeInstall, AfterInstall, StartApplication, ValidateService
EM_SERVICE_SLICE	The slice the service is being installed to. One of blue, green, none
EM_SERVICE_NAME	The name of the service being installed
EM_SERVICE_PORT	The port on which the service should listen
EM_SERVICE_VERSION	The version of the service being installed

Command line options

$ python agent/core.py -h
usage: core.py [-h] [-config-dir CONFIG_DIR] [-v]

optional arguments:
  -h, --help            show this help message and exit
  -config-dir CONFIG_DIR
                        Location of configuration files (e.g. config.yml and
                        config-logging.yml)
  -v, --version         show program's version number and exit

Configuration

Deployment agent supports multiple configuration files. The agent will look for configuration files in the same directory as the executable.

Filename	Description	Example
config-logging.yml	(optional) Python logging module configuration. If not specified, logs will be sent to stdout with level set to DEBUG.	[sample-config-logging.yml] (https://github.thetrainline.com/PlatformServices/consul-deployment-agent/blob/master/config/sample-config-logging.yml)
config.yml	(optional) Various configuration settings. See example for supported options.	[sample-config.yml] (https://github.thetrainline.com/PlatformServices/consul-deployment-agent/blob/master/config/sample-config.yml)

Development

Prerequisites (Linux and Windows)

Please make sure you have downloaded and setup the following first. You might also need to update your path environment variables for this to work.

1. Python 2.7.X
2. Consul agent

Setting up your PYTHONPATH

If you are checking this out for the first time you will need to make sure that you have setup your PYTHONPATH environment variable.

For example if you have checked out this project to:

C:\code\consul-deployment-agent

OR

~/code/consul-deployment-agent

Then you would need to make sure you set this exact path as a value.

For windows:

SET PYTHONPATH=C:\code\consul-deployment-agent

For linux:

export PYTHONPATH=~/code/consul-deployment-agent

Linux development

First, initialise the dependencies by running:

  sudo make init

Start Consul agent in development mode:

  consul agent -dev

To start the deployment agent:

  python agent/core.py

Running the tests:

  sudo make init-test
  make test

Windows development

First, initialise the dependencies by running:

  pip install -r requirements.txt

Start Consul agent in development mode:

  consul agent -dev -advertise=127.0.0.1

Start the deployment agent:

  python agent/core.py

Running the tests:

  pip install -r test-requirements.txt
  nosetests --verbosity=2 tests

Deployment simulation

Follow these steps to perform a deployment simulation:

1. Upload a deployment package to tl-deployment-sandbox S3 bucket in the Sandbox account
2. Start Consul agent as per instructions above.
3. Start Consul deployment agent as per instructions above.
4. Run the following script to trigger the deployment:

  ./scripts/deploy_service.py -n <service_name> -p <service_port> -v <version> -s <slice_name> -t <environment_type> -c <cluster_name> -b tl-deployment-sandbox -k <s3_object_key>