Monitoring a cluster

Follow a cluster through its Activity, its control-plane logs, its measured API availability and the metrics you can set alarms on.

Last updated 18 November 2026View as Markdown

Replace my-cluster with your cluster's name or id.

Activity

Every provisioning step, change, upgrade step, repair, snapshot and access change of a cluster is an event. Reading them needs eks:DescribeCluster.

  1. Open Kubernetes, open the cluster, then choose Activity.

Events come oldest first: without since, the first 200 (or limit, up to 500) since the cluster was created. To follow new events, ask again with since set to the last event's createdAt. Each event has a source (control_plane or machine), a level (info, warn, error), a phase and a message.

Problems on the cluster's page

The cluster's page shows one banner at a time: a problem the cluster reported ("The cluster reported a problem."), then a node group that stopped ("Node group workers stopped: no nodes are being created.") or is not getting its nodes ("Node group workers is not getting its nodes."), then the end of support of its Kubernetes minor. The node-group banner needs eks:ListNodegroups; without it, it is not shown.

Control-plane logs

The Kubernetes audit log and the control plane's own logs, kept 7 days; see Read control-plane logs.

Availability

Deutsche Cloud measures the API of every cluster that is active, updating or restoring once a minute; see Availability and the SLA.

Metrics and alarms

The cluster's Monitoring tab shows its per-minute metrics, and alarms can be set on them under the resource type eks_cluster. Reading the metrics needs monitoring:GetMetricData; the alarms, monitoring:DescribeAlarms.

Did this page help you?