Skip to main content
After the hub is running, install the exporter chart into each Kubernetes cluster you want to monitor. Each monitored cluster runs its own exporter, node agent, Redis, OpenTelemetry Collector, and target allocator. Those components collect local telemetry and send it outbound to the hub.
The install steps, endpoint values, and trust settings below use the decisions saved in Installation Planning.

Selected plan

1

Confirm prerequisites

Use a kube-context that points at the monitored cluster, not the hub cluster.
2

Create the namespace

3

Prepare exporter inputs

The exporter needs an environment token and the hub endpoints it should call.
The exporter bearer token is environment-name specific. If you change the cluster name, Metoro generates a different token. Use the token shown for the final cluster name.
Treat the exporter bearer token like a secret. Do not commit it, put it in the values file, paste it into shared tickets, or store it in broadly accessible shell history.
4

Create the values file

5

Install with Helm

6

Run post-install checks

7

Go back to Metoro UI

​
What Gets Installed

The exporter chart installs a cluster-local forwarding path. Metoro Exporter receives telemetry from the local collectors and node agents, watches Kubernetes resources, reads exporter configuration from the hub apiserver, and forwards telemetry to the hub ingester. Node Agent runs as a privileged DaemonSet on monitored nodes. It needs host access, host filesystem mounts, and container runtime socket mounts so it can collect node and workload telemetry. It is not suitable for serverless node environments that do not expose the host to pods. Redis is used for exporter coordination inside the monitored cluster. It does not store durable telemetry. If Redis state is lost, exporter-side maps are rebuilt as the exporter and node agents reconnect and observe the cluster again. When ServiceMonitor scraping is enabled, the chart also installs an OpenTelemetry Collector and Target Allocator. The monitored cluster must already have the Prometheus Operator ServiceMonitor and PodMonitor CRDs installed for those resources to be discovered. After the exporter connects, the monitored cluster should appear in the Metoro UI. It can take a few minutes for enough telemetry to arrive for services, workloads, and Kubernetes resources to populate.

​
Troubleshooting

If exporter pods cannot reach the hub, check outbound network policy, DNS, and the exact hub URLs in metoro-exporter-values.yaml. If pods fail with certificate errors and the hub uses private PKI, verify that the CA ConfigMap exists in the metoro namespace and that the values file references the same ConfigMap key. If node-agent pods are missing from some nodes, check node selectors, tolerations, privileged workload policy, and whether those nodes expose the host paths and runtime sockets required by the DaemonSet. If ServiceMonitor or PodMonitor targets are missing, confirm the Prometheus Operator CRDs exist in the monitored cluster and that the target allocator selectors match the resources you expect to scrape.