Generate Daily Reports From Metrics Using LLM

[songk1992]

Implement LLM-powered daily system health report generation

Description:

Create a pipeline that converts raw node, pod, and container metrics (MetricNodeEntity, MetricPodEntity, MetricContainerEntity) into a human-readable daily system health report using an LLM.

Requirements:

Aggregate metrics over a 24-hour period (CPU, memory, filesystem, network).

Compute derived values (percentages, deltas, peaks, anomalies).

Convert aggregates into an LLM-ready prompt.

Call the selected LLM provider to generate a natural-language report.

Output the report as Markdown for use in dashboards or notifications.

Include error handling for missing or partial metrics.

Acceptance Criteria:

Running the job produces a stable daily report summarizing node/pod/container health.

Reports include trends, anomalies, and capacity warnings.

The LLM receives structured input and returns a human-friendly summary.