Monitoring cluster state Managed Service for Kubernetes

Managed Service for Kubernetes automatically sends cluster metrics to Yandex Monitoring. Metrics are available for the following Kubernetes objects:

• Container
• Master
• Node
• Pod
• Persistent volume

You can find metrics description in Yandex Monitoring metric reference.

How to get statisticsHow to get statistics

There are several alternative ways:

• View the statistics in the management console. It visualizes only the metrics for master hosts, nodes, and pods.

• Open the Monitoring web interface. It visualizes all Managed Service for Kubernetes metrics.

• Upload metrics using the Monitoring API. This method is suitable if you want to get only numerical values of metrics without visualization.

• Install the Metrics Provider app. You can use it to transmit metrics in two directions:

  • From Kubernetes objects to monitoring systems and automatic scaling systems
  • From Monitoring to Kubernetes objects

• Install the Prometheus Operator app to work with the Yandex Managed Service for Prometheus® monitoring system. This way you can work with dashboards in Grafana and take advantage of Prometheus.

Viewing statistics in the management consoleViewing statistics in the management console

Viewing master statisticsViewing master statistics

Viewing node statisticsViewing node statistics

Viewing pod statisticsViewing pod statistics

Viewing statistics in MonitoringViewing statistics in Monitoring

1. Open the Monitoring home page.
2. In the left-hand panel, select Metric Explorer.
3. In the request field, click .
4. Configure a request for statistics.
5. Specify the statistics period.
6. Click Run.

Requests for statistics in MonitoringRequests for statistics in Monitoring

• For a Managed Service for Kubernetes container:

  "<metric>"{folderId="<folder_ID>", service="managed-kubernetes", cluster_id="<cluster_name>", container="<container_name>"}
  

  In the request, specify the following parameters:

  • "<metric>": Container metric name, e.g., "container.cpu.core_usage_time". The metrics are listed in the reference.
  • folderId: Folder ID.
  • cluster_id: Managed Service for Kubernetes cluster name.
  • container: Name of the container you need the statistics for.

• For a Managed Service for Kubernetes master:

  "<metric>"{folderId="<folder_ID>", service="managed-kubernetes", cluster_id="<cluster_name>"}
  

  In the request, specify the following parameters:

  • "<metric>": Container metric name, e.g., "master.cpu.utilization_percent". The metrics are listed in the reference.
  • folderId: Folder ID.
  • cluster_id: Managed Service for Kubernetes cluster name.

• For a Managed Service for Kubernetes node:

  "<metric>"{folderId="<folder_ID>", service="managed-kubernetes", cluster_id="<cluster_name>", node="<node_name>"}
  

  In the request, specify the following parameters:

  • "<metric>": Container metric name, e.g., "node.cpu.core_usage_time". The metrics are listed in the reference.
  • folderId: Folder ID.
  • cluster_id: Managed Service for Kubernetes cluster name.
  • node: Name of the node you need the statistics for.

• For a Managed Service for Kubernetes pod:

  "<metric>"{folderId="<folder_ID>", service="managed-kubernetes", cluster_id="<cluster_name>", pod="<pod_name>"}
  

  In the request, specify the following parameters:

  • "<metric>": Container metric name, e.g., "pod.memory.used_bytes". The metrics are listed in the reference.
  • folderId: Folder ID.
  • cluster_id: Managed Service for Kubernetes cluster name.
  • pod: Name of the pod you need the statistics for.

• For a Managed Service for Kubernetes persistent volume claim (PVC):

  "<metric>"{folderId="<folder_ID>", service="managed-kubernetes", cluster_id="<cluster_name>", persistentvolumeclaim="<PVC_name>"}
  

  In the request, specify the following parameters:

  • "<metric>": Container metric name, e.g., "pod.memory.used_bytes". The metrics are listed in the reference.
  • folderId: Folder ID.
  • cluster_id: Managed Service for Kubernetes cluster name.
  • persistentvolumeclaim: Name of the persistent volume you need the statistics for.