Observability

The portal includes a built-in, per-cluster observability view — you don’t set up a separate dashboard, install agents, or log in anywhere else. Open a cluster and the metrics are rendered natively right in the portal.

What you can see

The observability view shows the health of the parts of your cluster Aether manages and the nodes you run on, over time:

• Control-plane health — CPU and memory (usage against requests and limits), the same broken down per component (API server, controller manager, scheduler), container restarts, network throughput, and the number of control-plane pods running.
• Cluster datastore — CPU and memory of the managed datastore that backs your control plane, plus your cluster’s database size, transaction rate (commits vs rollbacks), and row activity.
• Worker nodes — per-node CPU, memory, network throughput, disk usage and provisioned disk size, and how many worker VMs are running.

Worker metrics are a machine-level view

Worker-node panels come from the hypervisor, not from inside your cluster. They show how loaded each node’s VM is — useful for capacity decisions — but they aren’t pod, container, or kubelet metrics. For those, run your own in-cluster monitoring.

How to read it

Open your cluster in the portal and select the observability view. Panels plot each metric over a time window you choose — 1 hour, 6 hours, 24 hours, or 7 days — so you can:

• Spot CPU or memory pressure building before it becomes a problem.
• Confirm a cluster is healthy after provisioning or a scaling change.
• Decide whether to scale a node pool up — or whether you have headroom to scale it down.

Note

The observability view focuses on cluster and node health metrics. For application-level logs, metrics, and traces of your own workloads, run your preferred in-cluster observability tooling alongside them.

Learn more

• Node pools — scale your nodes based on what you see here.