As a MLOps engineer I would like to have a collector of prometheus metrics at my edge node, so I can collect the operational metrics specific to my model.

[LaVLaS]

Sub-tasks

• Enable UWM (user workload monitoring ) on edge clusters and document that as part of the POC edge cluster(s) prerequisites. #12
• Identify model metrics that will be part of the PoC. #13
• Define the Prometheus rules and monitors to collect the documented PoC metrics. #14

A/C

• User workload monitoring is enabled on test edge clusters.
• The enabling of user workload monitoring is documented as part of the PoC edge cluster(s) prerequisites.
• Needed metrics for the PoC are documented.
• Prometheus rules and monitors are created on test edge clusters.
• The documented metrics can be queried on edge clusters using Prometheus expression browser.
• Prometheus rules and monitors are committed and pushed to the ai-edge repo.