Some system metrics are unavailable

Dashboard, Monitoring & Debugging
1

1. Severity of the issue: (select one)
High: Completely blocks me.

2. Environment:

  • Ray version: 2.46.0
  • Python version: 3.10.12
  • OS: linux kernel 5.10.134-16.101.al8.x86_64
  • Cloud/Infrastructure:
  • Other libs/tools (if relevant):

3. Repro steps / sample code: (optional, but helps a lot!)

Your example code

4. What happened vs. what you expected:

  • Expected: All system metrics are available
  • Actual: some metrics are unavailable

I started a Ray cluster using the command: ray start --head --port=6379 --metrics-export-port=8080 and with curl http://127.0.0.1:8080/metrics to fetch the system metrics. But I only got a small subset of them (e.g., ray_node_memory_available ). Many other metrics, such as ray_object* and ray_internal* , are missing.