4.x MP metrics and health do not honor server.features.observe.observers config, but instead observe.providers.*

[tjquinno]

Environment Details

• Helidon Version: 4.0.0
• Helidon SE or Helidon MP MP
• JDK version:
• OS:
• Docker version (if applicable):

---

Problem Description

In MP apps, metrics and health follow config at the top level (metrics and health) but not config inside server.features.observe.observers.metrics or ...health as SE apps do.

But they do respond to config at observe.providers.xxx.

This seems to be because the constructors for MetricsCdiExtension and HealthCdiExtension incorrectly pass hardcoded strings "observe.providers.metrics" and "observe.providers.health" to the superclass constructor as one of the config paths to consider.

I assume the intended behavior is for the CDI extensions to use either the top-level config for metrics and health or the config nested under server... but the paths do not have the correct prefix and use providers instead of observers.

Steps to reproduce

1. Build and run the MP QuickStart app without change, then run this command: curl -v http://localhost:8080/health. You get the expected output with status 200

   {"status":"UP","checks":[]}

   because the MP health CDI extension sets the health setting details to true programmatically.
2. Stop the server, then add the following config:

   server.features.observe.observers.health.details = false

3. Rebuild and rerun the server. The output is the same because the health CDI observer ignores the config under server.features....