18f / analytics-restarter

Triggers nightly builds for the apps that makeup the analytics.usa.gov system

Analytics Restarter

This app triggers nightly builds for the apps that makeup the analytics.usa.gov system. These apps include:

• analytics.usa.gov
• analytics-reporter
• analytics-reporter-api

This service will restart CI builds for the above apps at 06:00 UTC every day.

Deploy

To setup this app in cloud.gov, first deploy a user provided service to provide the app with the necessary information to start builds on circle. The app looks for a user provided service named circleci. It expects the service to have a key for each app referencing a JSON encoded string providing the repo name and a CircleCI token for the app.

For example:

$ cf cups -p ANALYTICS_USA_GOV,ANALYTICS_REPORTER,ANALYTICS_REPORTER_API

ANALYTICS_USA_GOV> {"NAME": "analytics.usa.gov","TOKEN": "123abc"}
ANALYTICS_REPORTER> {"NAME": "analytics-reporter","TOKEN": "456def"}
ANALYTICS_REPORTER_API> {"NAME": "analytics-reporter-api","TOKEN": "789ghi"}

This app also provides a route at / to provide an endpoint for New Relic uptime monitoring.

Tests

This app ships with a lightweight testing suite. That can be run with rspec:

bundle install
bundle exec rspec

analytics-restarter's Issues

Setup circle

We need to add this app to CircleCI so that it can restart itself.

How difficult is this to modularize?

Like, if other 18F (or other cloud.gov customer) projects wanted to use this to satisfy the same control?

Add reporting for failed builds

If the app encounters an error and fails to start a new build for a given project, we want the restarter to have a means by which to alert maintainers that the restart has failed. This could be something provided in the environment, or could potentially pull the contact info from the circle or GitHub api.