The Full Stack dashboard provides job-aware observability by correlating physical infrastructure health with specific job workloads, so you can determine whether a performance dip is caused by underlying hardware issues.
Dashboard Views
-
AI Cluster Health: A high-level overview of the entire environment's operational status.
-
NVIDIA Minerva Ingestion: Monitors the health and latency of the telemetry data pipeline.
-
Job Failure Root Cause Analysis: A specialized view for debugging jobs by mapping telemetry events to the job's lifecycle.
-
Telemetry Health: Ensures that all telemetry sources (switches, NICs, hosts, and other sources) across the stack are reporting correctly.
-
North-South AI Cluster Health: Monitors traffic entering and exiting the cluster, along with interface health indicators for external-facing network paths.
-
Fabric Validation: Monitors fabric health across AI factory network infrastructure.
Last updated: