SambaNova Hardware Monitoring gives your team complete visibility into RDU performance, utilization, and health across every node in the cluster – including RDU health status, utilization percentages, high-level error and fault counts, host BMC temperatures, fan speeds, voltages, and XRDU power and thermal data.The monitoring stack is built on three Prometheus exporters (IPMI, SambaNova, and XRDU), deployed as Kubernetes DaemonSets on your cluster. Each exporter collects hardware metrics from a different layer of your SambaRack infrastructure.Optionally, the chart can install a kube-prometheus-stack ServiceMonitor CRD to automatically scrape metrics using Prometheus. Alternatively, customers can configure their own monitoring directly from the exporter metrics endpoints.
For custom use cases or direct exporter configuration, contact SambaNova support.
SambaRack Manager (snctl) is a prerequisite. You will use snctl to look up the BMC and XRDU IP addresses needed during installation.
Ready to install? Start with prerequisites.
Prerequisites
What you need before installing the hardware monitoring stack
Optionally, the chart can install a kube-prometheus-stack ServiceMonitor CRD to automatically scrape metrics using Prometheus. Alternatively, customers can configure their own monitoring directly from the exporter metrics endpoints.
For custom use cases or direct exporter configuration, contact SambaNova support.
SambaRack Manager (snctl) is a prerequisite. You will use snctl to look up the BMC and XRDU IP addresses needed during installation.
