Metrics collected by Container insights

Container insights collects custom metrics from Azure Kubernetes Service (AKS) and Azure Arc-enabled Kubernetes cluster nodes and pods. With custom metrics, you can:

• Present timely aggregate calculations (average, count, maximum, minimum, and sum) in performance charts.
• Pin performance charts in Azure portal dashboards.
• Take advantage of metric alerts.

Note

This article describes collection of custom metrics from Kubernetes clusters. You can also collect Prometheus metrics as described in Collect Prometheus metrics with Container insights.

Use custom metrics

Custom metrics collected by Container insights can be accessed with the same methods as custom metrics collected from other data sources, including metrics explorer and metrics alerts.

Metrics collected

The following sections describe the metric values collected for your cluster.

Node metrics

Namespace: Insights.container/nodes
Dimensions: host

Metric	Description
cpuUsageMillicores	CPU utilization in millicores by host.
cpuUsagePercentage, cpuUsageAllocatablePercentage (preview)	CPU usage percentage by node and allocatable, respectively.
memoryRssBytes	Memory RSS utilization in bytes by host.
memoryRssPercentage, memoryRssAllocatablePercentage (preview)	Memory RSS usage percentage by host and allocatable, respectively.
memoryWorkingSetBytes	Memory Working Set utilization in bytes by host.
memoryWorkingSetPercentage, memoryRssAllocatablePercentage (preview)	Memory Working Set usage percentage by host and allocatable, respectively.
nodesCount	Count of nodes by status.
diskUsedPercentage	Percentage of disk used on the node by device.

Pod metrics

Namespace: Insights.container/pods
Dimensions: controllerName, Kubernetes namespace

Metric	Description
podCount	Count of pods by controller, namespace, node, and phase.
completedJobsCount	Completed jobs count older user configurable threshold (default is six hours) by controller, Kubernetes namespace.
restartingContainerCount	Count of container restarts by controller and Kubernetes namespace.
oomKilledContainerCount	Count of OOMkilled containers by controller and Kubernetes namespace.
podReadyPercentage	Percentage of pods in ready state by controller and Kubernetes namespace.

Container metrics

Namespace: Insights.container/containers
Dimensions: containerName, controllerName, Kubernetes namespace, podName

Metric	Description
(Old)cpuExceededPercentage	CPU utilization percentage for containers exceeding user configurable threshold (default is 95.0) by container name, controller name, Kubernetes namespace, and pod name. Collected
(New)cpuThresholdViolated	Metric triggered when CPU utilization percentage for containers exceeding user configurable threshold (default is 95.0) by container name, controller name, Kubernetes namespace, and pod name. Collected
(Old)memoryRssExceededPercentage	Memory RSS percentage for containers exceeding user configurable threshold (default is 95.0) by container name, controller name, Kubernetes namespace, and pod name.
(New)memoryRssThresholdViolated	Metric triggered when Memory RSS percentage for containers exceeding user configurable threshold (default is 95.0) by container name, controller name, Kubernetes namespace, and pod name.
(Old)memoryWorkingSetExceededPercentage	Memory Working Set percentage for containers exceeding user configurable threshold (default is 95.0) by container name, controller name, Kubernetes namespace, and pod name.
(New)memoryWorkingSetThresholdViolated	Metric triggered when Memory Working Set percentage for containers exceeding user configurable threshold (default is 95.0) by container name, controller name, Kubernetes namespace, and pod name.

Persistent volume metrics

Namespace: Insights.container/persistentvolumes
Dimensions: kubernetesNamespace, node, podName, volumeName

Metric	Description
(Old)pvUsageExceededPercentage	Persistent volume (PV) utilization percentage for persistent volumes exceeding user configurable threshold (default is 60.0) by claim name, Kubernetes namespace, volume name, pod name, and node name.
(New)pvUsageThresholdViolated	Metric triggered when PV utilization percentage for persistent volumes exceeding user configurable threshold (default is 60.0) by claim name, Kubernetes namespace, volume name, pod name, and node name.

Enable custom metrics

If your cluster uses managed identity authentication for Container insights, custom metrics will be enabled for you. If not, you need to enable custom metrics by using one of the following methods.

This process assigns the Monitoring Metrics Publisher role to the cluster's service principal. Monitoring Metrics Publisher has permission only to push metrics to the resource. It can't alter any state, update the resource, or read any data. For more information, see Monitoring Metrics Publisher role. The Monitoring Metrics Publisher role requirement doesn't apply to Azure Arc-enabled Kubernetes clusters.

Prerequisites

Before you update your cluster:

• See the supported regions for custom metrics at Supported regions.
• Confirm that you're a member of the Owner role on the AKS cluster resource to enable collection of custom performance metrics for nodes and pods. This requirement doesn't apply to Azure Arc-enabled Kubernetes clusters.

Enablement options

Use one of the following methods to enable custom metrics for either a single cluster or all clusters in your subscription.

Azure portal

1. Select the Insights menu for the cluster in the Azure portal.

2. On the banner that appears at the top of the pane, select Enable to start the update.

   

   The process can take several seconds to finish. You can track its progress under Notifications from the menu.

CLI

Update a single cluster

In the following command, edit the values for subscriptionId, resourceGroupName, and clusterName by using the values on the AKS Overview page for the AKS cluster. The value of clientIdOfSPN is returned when you run the command az aks show.

az login
az account set --subscription "<subscriptionName>"
az aks show -g <resourceGroupName> -n <clusterName> --query "servicePrincipalProfile"
az aks show -g <resourceGroupName> -n <clusterName> --query "addonProfiles.omsagent.identity"
az role assignment create --assignee <clientIdOfSPN> --scope <clusterResourceId> --role "Monitoring Metrics Publisher" 

To get the value for clientIdOfSPNOrMsi, run the command az aks show as shown in the following example. If the servicePrincipalProfile object has a valid objectid value, you can use that. Otherwise, if it's set to msi, pass in the object ID from addonProfiles.omsagent.identity.objectId.

az login
az account set --subscription "<subscriptionName>"
az aks show -g <resourceGroupName> -n <clusterName> --query "servicePrincipalProfile"
az aks show -g <resourceGroupName> -n <clusterName> --query "addonProfiles.omsagent.identity" 
az role assignment create --assignee <clientIdOfSPNOrMsi> --scope <clusterResourceId> --role "Monitoring Metrics Publisher"

Note

If you want to perform the role assignment with your user account, use the --assignee parameter as shown in the example. If you want to perform the role assignment with a service principal name (SPN), use the --assignee-object-id and --assignee-principal-type parameters instead of the --assignee parameter.

Update all clusters

Run the following command to update all clusters in your subscription. Edit the value for subscriptionId by using the value on the AKS Overview page for the AKS cluster.

az login
az account set --subscription "Subscription Name"
curl -sL https://aka.ms/ci-md-onboard-atscale | bash -s subscriptionId   

The configuration change can take a few seconds to finish. When it's finished, a message like the following one appears and includes the result:

completed role assignments for all AKS clusters in subscription: <subscriptionId>

PowerShell

Update a single cluster

To enable custom metrics for a specific cluster:

1. Download the mdm_onboarding.ps1 script from GitHub and save it to a local folder.

2. Run the following command. Edit the values for subscriptionId, resourceGroupName, and clusterName by using the values on the AKS Overview page for the AKS cluster.

   .\mdm_onboarding.ps1 subscriptionId <subscriptionId> resourceGroupName <resourceGroupName> clusterName <clusterName>
   

   The configuration change can take a few seconds to finish. When it's finished, a message like the following one appears and includes the result:

   Successfully added Monitoring Metrics Publisher role assignment to cluster : <clusterName>
   

Update all clusters

To enable custom metrics for all clusters in your subscription:

1. Download the mdm_onboarding_atscale.ps1 script from GitHub and save it to a local folder.

2. Run the following command. Edit the value for subscriptionId by using the value on the AKS Overview page for the AKS cluster.

   .\mdm_onboarding_atscale.ps1 subscriptionId
   

   The configuration change can take a few seconds to finish. When it's finished, a message like the following one appears and includes the result:

   Completed adding role assignment for the aks clusters in subscriptionId :<subscriptionId>
   

Verify the update

To verify that custom metrics are enabled, open metrics explorer and verify from Metric namespace that insights is listed.

Next steps

• Create alerts based on custom metrics collected for the cluster
• Collect Prometheus metrics from your AKS cluster