Recommendations Broker

The recommendations-broker collects recommendations from third-party services every so often and sends them to the controller-api with proper formatting.

Usage

Environment

The following environment variables need to be set:

Environment Variable	Description	Required	Example
AKKERIS_APP_CONTROLLER	Endpoint for the Akkeris api	yes	https://apps.akkeris.io
AKKERIS_SERVICE_TOKEN	Service token used for authenticating to Akkeris	yes	ey...
LOG_LEVEL	Change the default log level (debug) for more or less information	no	info

Turbonomic

Environment Variable	Description	Required	Example
TURBO_USERNAME	Turbonomic username	yes	svc-turbo-user
TURBO_PASSWORD	Turbonomic password	yes	hunter2
TURBO_URL	URL of the Turbonomic instance	yes	https://turbo.akkeris.io
KUBERNETES_CLUSTER_NAME	Target Kubernetes cluster (currently only one supported)	yes	ds1
AKKERIS_STACK_NAME	Target Akkeris stack (currently only one supported)	yes	ds1

Running

npm install
npm start

Or:

docker build -t recommendations-broker .
docker run --rm --name recommendations-broker -p 9000:9000 recommendations-broker

Note:

The DigiCertSHA2HighAssuranceServerCA.pem file has been added to the node.js EXTRA_CA_CERTS environment variable in the start script because for some reason Turbonomic was not including it as an intermediary for our Turbonomic instance. YMMV

TODO

Turbonomic

• Search for apps in multiple clusters

Contributing

Code Structure

The recommendations-broker is designed to be pluggable - it will run any job that is placed in the services folder provided that it follows a common convention (see Adding a New Service below)

The index.js entrypoint loads the web server (server/index.js, provides basic healthcheck / loaded service information) and all of the jobs in the services folder.

Helpful functions have been provided in the utils.js file, and are mostly functions to communicate with Akkeris.

Adding a new service

This project uses a job scheduler called Bree.js to run tasks.

To add a new service that provides recommendations, create a new folder under the services directory with an index.js file exporting the following:

• An init function - can be used to set up services if needed
• A service string - containing the name of the service
• A jobs array - an array of objects consisting of:
  • name - Name of the job
  • filename - Filename of the job
  • interval - How often the job should run

The individual job files (referenced in the jobs array) should be placed in a jobs folder, which should be nested under the new service folder in the services directory.

For an example of this structure, take a look at the services/turbonomic directory.