Observability agents - Nebius AI Cloud

Nebius AI Cloud Observability uses two agents for different purposes:

Monitoring agent collects system metrics from virtual machines and Managed Service for Kubernetes® nodes for visualization on preconfigured dashboards. It can also collect journald logs from VMs.
Nebius Observability Agent for Kubernetes ingests logs from Managed Service for Kubernetes clusters into Observability.

See the detailed comparison of the two agents below:

Feature	Nebius Observability Agent for Kubernetes	Monitoring agent
Purpose	Ingests logs from Managed Service for Kubernetes clusters into Observability	Collects system-level metrics (CPU, RAM, GPU, etc.) and journald logs from Compute VMs
Deployed on	Managed Service for Kubernetes clusters	Compute virtual machines and Managed Service for Kubernetes nodes
Collected data	Application and infrastructure logs	System metrics (for example, GPU, vCPU, memory and OS) and journald logs
Installation	Manual (via Helm chart)	Automatic (preinstalled on new VMs and Managed Kubernetes nodes)
Components	Helm chart with the log collection agent	`nebius-observability-agent`, `nebius-observability-agent-updater`
Data enrichment	Adds metadata: cluster ID, pod name, container image tag, etc.	Adds labels: resource ID, project ID, etc.
Data destination	Observability Logs	Observability Metrics (web console dashboards)
Management	Version controlled via Helm and customizable installation	Managed via SSH, supports disabling updates
Use case	Centralized log ingestion for observability and troubleshooting	Visual monitoring of infrastructure resource usage