hasura-metrics-demo

Spring boot application that pulls metrics data from Hasura Pro Metrics Service and notifies a downstream client (eg. Slack, etc,.)

Pre-requisites

• Access to a Hasura Pro Backend (LUX) service, especially the Metrics service URL and its admin-secret.
• Slack App/Channel to post the error messages and User/Bot token to authenticate with the Slack API (Reference) having correct scopes.

Running the application

• Add the Hasura Metrics service Graphql API URl, Metrics servcie admin secret, Slack USer/Bot token and Slack channel/App id to application.properties file.
• Build and run the Spring boot application.

How it works!

• The class ErrorFetchingService.java has the funtion that queries the metrics API to get the errors that occurred in the last 10 seconds.
• The class ErrorNotificationJob.java has the scheduler that runs the above query every 10 seconds (Configurable in code). For Non empty responses , it calls the notify method in SlackNotifier.java to post the error to Slack.

Extending for other metrics

• The metrics service of LUX is itself a Hasura graphql engine and the Hasura console is exposed (by default) on port 8080 of the container/pod. The graphiql view of metrics service is self-documenting about the metrics exposed as graphql-apis. They can be consumed just like any other Graphql resources from other applications.
  • An easy way to access the metrics service's console is to use kubectl port-forward (Reference). To map local port 8090 to port 8080 of the metrics pod (by default has name matching the wildcard metrics-*, for example metrics-54ffd6d96d-k5c6h), use kubectl port-forward metrics-54ffd6d96d-k5c6h 8090:8080 . Now the Hasura console for metrics is availble at http://localhost:8090 .