Open
Description
Summary
Operators responsible for maintaining newrelic daemons only have logs for visibility into problems with the daemon. Adding a metric endpoint would allow for operators to more easily identify problems with data collection.
Desired Behavior
The desired behavior is a prometheus/openmetrics endpoint that is accessible on every newrelic daemon that can be queried for metrics and exported to metric aggregation services using normal prometheus tooling.
This will allow operators to quickly debug issues and configure Horizontal Pod Autoscalers to scale on certain metrics in Kubernetes environments.
The following metrics would be useful
- applications_connected # can be used by an HPA to schedule additional newrelic pods
- agents_connected # can be used to determine how many pods are connected
- request_total # can be used to calculate requests per second
- request_errors_ total # can be used to determine how many requests were not sent to api.newrelic.com
- request_sent_kilobytes_total # can be used to calculate bytes per second
Possible Solution
The is can be accomplished with prometheus or opentelemetry sdks