[OpenShift Metrics Access Request] Metrics and logs access for GPU/MIG validation

[djnakabaale]

Please answer some questions to justify granting your team or user access to NERC OpenShift metrics and logs.

Do you need to be able to view the application/infrastructure/audit logs of the OpenShift clusters?

• Application logs: Yes (required for operator debugging/validation).
• Infrastructure logs: Prefer read access if possible for troubleshooting metrics pipeline issues.
• Audit logs: Not required for this work.

Do you need user access to view NERC OpenShift metrics in Grafana for certain clusters or all clusters?

• Yes, for the specific cluster(s) and namespace(s) associated with my project/testing only (not all clusters).

Are you building a microservice to query the metrics?

• No. I am validating an existing Kubernetes/OpenShift operator that queries Prometheus metrics from within OpenShift.

Are you pulling metrics once by hand?

• Yes, for ad-hoc validation and troubleshooting during development.

Are you pulling them on a schedule?

• Yes. The operator performs scheduled collection/reporting during runtime.

How long will you need to query the metrics?

• Temporary access for active development/validation, approximately up to 4 weeks altogether

How much data do you plan to collect?

• Limited to project validation scope (GPU/MIG and supporting workload/cluster metrics needed by the operator)
• Expected volume is modest and tied to operator report generation/testing only

Which metrics do you need to observe?

• NVIDIA/DCGM GPU metrics (including MIG-related series)
• kube-state-metrics/pod/node metadata
• Prometheus query/results behavior needed to validate generated reports

Also, NERC metrics data must remain in in NERC systems only. Were you planning on pulling metrics out of NERC for any reason (research data gathering, reports)?

• No. Metrics data will remain within NERC systems.
• Access is for in-cluster validation and troubleshooting.
• Any externally shared material will be sanitized summaries, not raw NERC metrics exports.

What is your NERC OpenShift project allocation name?

• GPU metering for Cost Management
• koku-metrics-operator

[djnakabaale]

cc: @rstahl-rh

[computate]

Access for the RH Cost Management On Prem team to the nerc-ocp-test cluster is done.

[joachimweyl]

What are the next steps? Can this be closed?

[computate]

This has been completed for Cost Management On Prem testing in MOC 1.0.

[hpdempsey]

Where are the results? We would like to see this, since we are looking at Cost Management for MOC 2.0