Skip to main content
Node Agent runs on monitored Kubernetes nodes and collects workload telemetry that requires host-level visibility. It is deployed as a DaemonSet so each eligible node can run a local collector.

​
Talks To

​
How It Fits

Node Agent is responsible for telemetry that needs direct node access. The chart runs it with host PID access, privileged security context, and host path mounts for cgroups, tracing, debugfs, kubelet data, and container runtime sockets. The agent sends traces, logs, and profiles to the Metoro Exporter service inside the monitored cluster. It also uses Redis for coordination data shared across exporter components. Because it is a DaemonSet, Node Agent capacity follows the node count. Adding nodes adds more agent pods; removing nodes removes the corresponding collectors.

​
Operational Notes

Node Agent requires host access and cannot run on environments that hide the host from pods, such as Fargate-style serverless nodes. Operators should make sure scheduling rules, tolerations, and security policy allow the DaemonSet to run on the nodes that should be monitored. If Node Agent is unhealthy on a node, telemetry from workloads on that node can be incomplete even when the hub and exporter Deployment are healthy.