Microsoft Azure Kubernetes Service (AKS) Monitoring

Overview

Azure Kubernetes Service (AKS) offers server-less Kubernetes, an integrated continuous integration and continuous delivery (CI/CD) experience and enterprise-grade security and governance. Azure Kubernetes Service (AKS) simplifies deploying a managed Kubernetes cluster in Azure by offloading the operational overhead to Azure. As a hosted Kubernetes service, Azure handles critical tasks, like health monitoring and maintenance.

Since Kubernetes masters are managed by Azure, you only manage and maintain the agent nodes. Thus, AKS is free; you only pay for the agent nodes within your clusters, not for the masters. It allows you to rapidly build, deliver and scale applications. Applications Manager's Azure Kubernetes Service monitoring tool provides visibility into the health, performance, and behavior of Kubernetes clusters, enabling administrators to ensure the reliability and efficiency of containerized applications running on AKS.

Creating a new Microsoft Azure Kubernetes Service monitor

To learn how to create a new Microsoft Azure Kubernetes Service monitor click here.

Monitored Parameters

Navigate to the Category View by clicking the Monitors tab. Hover over 'Child Monitors' under Microsoft Azure in the Cloud Apps table, and then select the Kubernetes Service monitor from the displayed tooltip. This action will display the bulk configuration view for Azure Kubernetes Service in three tabs:

• Availability tab gives the availability history for the past 24 hours or 30 days.
• Performance tab gives the health status and events for the past 24 hours or 30 days.
• List view tab enables you to perform bulk admin configurations.

The Microsoft Azure monitor provides a brief detail of the Azure Kubernetes Service under the given subscription. Following are the list of metrics monitored in Azure Kubernetes Service Monitoring in their corresponding tabs:

• Overview
• Node Pool
• Node
• Pods
• Services
• Persistent Volumes
• Service Map

Overview

Parameter	Description	Data Collection Method
		Kubectl	Azure Monitor
CLUSTER INFORMATION
Status	Current status of the cluster. (Running, Stopped)
Cluster FQDN	The Azure Resource Name (ARN) of the cluster.
Cluster Location	Location where the cluster resides.
Cluster Kubernetes Version	Kubernetes version of the cluster server.
CLUSTER DETAILS
Git Version	Git version of the cluster.
Build Date	Build date of the cluster.
Compiler	Name of the compiler used.
Platform	OS Platform of the cluster.
CLUSTER AUTOSCALER DETAILS (Applicable only when cluster autoscaler is enabled)
Cluster Scaledown Cooldown	Determines if the scale down is in cooldown. No nodes will be removed during this timeframe at the time of polling.
Cluster UnNeeded Nodes	Cluster autoscaler marks those nodes as candidates for deletion and are eventually deleted at the time of polling
Cluster UnSchedulable Pods	Number of pods that are currently unschedulable in the cluster at the time of polling.
Cluster Health	Determines the health of the autoscaler at the time of polling.
NAMESPACE DETAILS
Namespace UUID	UUID of the namespace.
Namespace Name	Name of the namespace
Pods Count	Number of pods count in the namespace at the time of polling.
Running pods in namespace	Number of running pods under the namespace at the time of polling.
Pending Pods	Number of pending pods under the namespace at the time of polling.
Succeeded Pods	Number of succeeded pods under the namespace at the time of polling.
Failed Pods	Number of failed pods under the namespace at the time of polling.
Unknown Pods	Number of succeeded pods under the namespace at the time of polling.
Restarting Container Count	Number of containers restarted count in the namespace at the time of polling.
OOMKilled Container Count	Total number of OOMKilled containers by namespace.
Pods Ready	Percentage of the number of pods that are in ready stage at the time of polling.
Namespace Resource Version	The version number of Namespace
Namespace Availability	Availability of Namespace
Namespace Created Time	Timestamp at which the Namespace was created
CLUSTER CPU & MEMORY DETAILS
Available Cluster CPU Cores	Total number of available CPU cores in the cluster at the time of polling.
Available Cluster Memory	Total number of available Memory in the cluster at the time of polling.
NODE DETAILS
Master Nodes	Number of master nodes available in the cluster.
Worker Nodes	Number of worker nodes available in the cluster.
CLUSTER USAGE DETAILS
Avg. CPU Usage	Average amount of CPU used by the cluster (in %).
Avg. Memory Usage	Average amount of memory used by the cluster (in %).
NODE COUNT DETAILS
Nodes Ready	Number of nodes that are in ready stage at the time of polling.
Nodes Not Ready	Number of nodes that are not in ready stage at the time of polling.
Top 5 Nodes by Used Pods - Displays a graphical representation of top 5 nodes with respect to the number of pods used.
PODS
Used pods in cluster %	Percentage of the number of pods used.
Used pods in cluster	Number of pods used in the cluster.
Maximum pods in cluster	Maximum number of pods available in the cluster.
Pods Ready in cluster	The average number of pods in the ready state at the time of polling.
PODS PHASE
Total Failed Pods	Total number of pods failed in the cluster at the time of polling.
Total Pending Pods	Total number of pods pending in the cluster at the time of polling.
Total Running Pods	Total number of pods running in the cluster at the time of polling.
Total Unknown Pods	Total number of pods unknown in the cluster at the time of polling.
Total Succeeded Pods	Total number of pods succeeded in the cluster at the time of polling.
API SERVER INFLIGHT REQUESTS
API Server Inflight Requests	The average of the maximum number of currently used inflight requests on the API server between the poll interval.
API SERVER USAGE (%)
API Server CPU Usage	The average of maximum CPU percentage (based off current limit) used by the API server pod across instances at the time of polling (in %).
API Server Memory Usage	The average of maximum memory percentage (based off current limit) used by the API server pod across instances at the time of polling (in %).
ETCD USAGE (%)
ETCD CPU Usage	The average of maximum CPU percentage (based off current limit) used by the ETCD pod across instances at the time of polling (in %).
ETCD Memory Usage	The average of maximum memory percentage (based off current limit) used by the ETCD pod across instances at the time of polling (in %).
ETCD Database Usage	The average of maximum utilization percentage of the ETCD database across instances at the time of polling (in %).

* Note: API SERVER INFLIGHT REQUESTS, API SERVER USAGE (%) and ETCD USAGE (%) group metrics are mapped under Performance Polling. To adjust the polling interval, navigate to Settings → Performance Polling. In the Optimize Data Collection tab, select Azure Kubernetes Service as the Monitor Type and Component Metrics as the Metric Name. Then set Default Polling Status as required, by default opted as Never collect data.

Node Pool

Parameter	Description	Data Collection Method
		Kubectl	Azure Monitor
Node Pool Details *
Name	Name of the node pool.
Status	The current status of the managed node pool. Possible values: Running/Stopped.
VM Size	The size of the agent pool VMs. VM size availability varies by region.
Kubernetes Version	The version of Kubernetes running on the agent pool.
Image Version	The version of node pool image.
Minimum Node Size (Auto-Scaling)	The minimum number of nodes for auto-scaling.
Maximum Node Size (Auto-Scaling)	The maximum number of nodes for auto-scaling.
Maximum Pods Count	The maximum number of pods that can run on a node.
Node Pool Utilization *
Name	Name of the node pool.
OS Type	The operating system type. Default value: Linux.
OS SKU	Specifies an OS SKU. This value must not be specified if OS type is Windows.
OS Disk Type	The OS disk type to be used for machines in the agent pool.
OS Disk Size	Size of the OS Disk to be used to specify the disk size for every machine in the master/agent pool (in GB).
CPU Usage	Average CPU utilization measured across the node pool over a specified poll interval (in %).
Disk Usage	Disk space used for node pool over a specified poll interval (in %).
Memory RSS Usage	Container RSS memory used for node pool over a specified poll interval (in %).
Memory WS Usage	Container working set memory used for node pool over a specified poll interval (in %).

* Note: Azure Monitor metrics are not available for Windows node pool.

Node

Parameter	Description	Data Collection Method
		Kubectl	Azure Monitor
Top 5 Nodes by Memory Details - Displays a graphical representation of top 5 nodes with respect to their corresponding memory details (in %).
Top 5 Nodes by CPU Details - Displays a graphical representation of top 5 nodes with respect to their corresponding CPU details (in %).
Node Usage Details *
Node Name	Name of the node.
CPU Usage Millicores	Average CPU utilization in millicores across the node over a specified poll interval.
CPU Usage (%)	Average CPU utilization measured across the node over a specified poll interval (in %).
Disk Usage (GB)	Disk space used for node over a specified poll interval (in GB).
Disk Usage (%)	Disk space used for node over a specified poll interval (in %).
Memory RSS Usage (GB)	Container RSS memory used for node over a specified poll interval (in GB).
Memory RSS Usage (%)	Container RSS memory used for node over a specified poll interval (in %).
Memory WS Usage (GB)	Container working set memory used for node over a specified poll interval (in GB).
Memory WS Usage (%)	Container working set memory used for node over a specified poll interval (in %).
Network In Usage	Amount of data received by the network for the node over a specified poll interval (in MB).
Network Out Usage	Amount of data transmitted by the network for the node over a specified poll interval (in MB).
Node Pod Details
Node Name	Name of the node.
Pod Usage	Number of pods available for usage with used and free pods split-up.
Used Pods	Number of pods that are used in the node.
Free Pods	Number of pods that are free in the node.
Allocatable Pods	Number of pods that are allocatable in the node.
Kube-system Pod Count	Number of Kube state pods in the node.
Non-Kube-system Pod Count	Number of non-Kube state pods in the node.
Images	Number of images present in the node.
NODE DETAILS
Node Name	Name of the node.
OS Image	OS Image name of the node.
OS	Name of the OS in which the container is deployed.
Architecture	Architecture details of the node.
Type	Type of node used.
Kubelet Version	The version of Kubelet used.
Allocatable Ephemeral Storage	Size of temporary memory available (in GiB).
Created Time	Timestamp at which the node was created.

* Note: Azure Monitor metrics are not available for Windows node.

Pods

Parameter	Description	Data Collection Method
		Kubectl	Azure Monitor
Pod Details
Pod UUID	Universal unique ID of the pod.
Pod Name	Name of the pod.
Pod Namespace	Namespace in which the pod resides.
Pod CPU Used (millicores)	The amount of CPU consumed by the pod (in millicores).
Pod CPU Usage (%)	The percentage of CPU utilized by the pod.
Pod Memory Used (MiB)	The amount of memory used by the pod (in MiB).
Pod Memory Usage (%)	The percentage of memory utilized by the pod.
Pod Node Name	Name of the pod-node.
Pod Application	Name of the pod application.
Pod Type	Type of pod used.
Pod created	Medium by which the pod was created.
Pod Status	Status of the pod. Possible values: Pending Running Succeeded Failed Unknown
Pod Age	Amount of time elapsed since the pod was created (in days).
Pod Start Time	Timestamp at which the pod was started.
Pod Created Time	Timestamp at which the pod was created.
Top 10 Pods by Memory Details - Displays a graphical representation of top 10 pods with respect to their corresponding memory details (in %).
Top 10 Pods by CPU Details - Displays a graphical representation of top 10 pods with respect to their corresponding CPU details (in %).
Pod Usage Details
Pod UUID	Universal unique ID of the pod.
Pod Name	Name of the pod.
Pod Namespace	Name of the pod.
Pod Containers Count	The number of containers run by the pod.
Pod CPU Limit(millicores)	The maximum amount of CPU resources that all containers in a Pod are allowed to use collectively (in millicores). If 0, no limit is set, then pod can use CPU upto the node capacity.
Pod CPU Limit(%)	The maximum amount of CPU resources that all containers in a Pod are allowed to use collectively (in %). It is the sum of the CPU limits set on each container within the Pod. The Kubernetes kubelet enforces this limit to ensure the Pod does not exceed the specified CPU usage on the node. If a container tries to use more CPU than its limit, it will be throttled.
Pod CPU Request (millicores)	The guaranteed minimum amount of CPU resources that all containers in a Pod collectively request (in millicores). If 0, no CPU is guaranteed and it can use CPU only if the node has free capacity.
Pod CPU Request(%)	The guaranteed minimum amount of CPU resources that all containers in a Pod collectively request (in %). It is the sum of the CPU requests of all containers in the Pod. Kubernetes uses this value to schedule the Pod onto a node that has enough CPU capacity to meet the request.
Pod Memory Limit(in MiB)	The maximum amount of memory (RAM) that all containers in a Pod can use collectively (in MiB).If 0, no limit is set, then pod can use memory upto the node capacity.
Pod Memory Limit(in %)	The maximum amount of memory (RAM) that all containers in a Pod can use collectively (in %). It is the sum of the memory limits of all containers in the Pod. The kubelet enforces this limit; if a container exceeds its memory limit, it may be terminated.
Pod Memory Request (in MiB)	The minimum amount of memory that all containers in a Pod request collectively (in MiB). If 0, no CPU is guaranteed and it can use memory only if the node has free capacity.
Pod Memory Request (in %)	The minimum amount of memory that all containers in a Pod request collectively (in %). It is the sum of the memory requests of all containers in the Pod. Kubernetes uses this value to schedule the Pod onto a node that has enough available memory to satisfy the request..
Pod created	Medium by which the pod was created.
Pod Persistent Volumes Claim	Name of the Claim through which a pod can access the persistent volume.
Container Details
Container ID	ID of the container.
Container Name	Name of the container.
Container Image	Name of the container image.
Container Pod Name	Name of the container pod.
Container Restarts	The number of times the container has restarted.
Container Status	Status of the container. Following are the list of possible values that are shown for each status: Status Value Running Running Waiting ContainerCreating CrashLoopBackOff ErrImagePull ImagePullBackOff CreateContainerConfigError InvalidImageName CreateContainerError Terminated OOMKilled Error Completed ContainerCannotRun DeadlineExceeded
Container Age	Amount of time elapsed since the container was started (in days).
Container Start Time	Timestamp at which the container was started.

Services

Parameter	Description	Data Collection Method
		Kubectl	Azure Monitor
Service Details
Service UUID	Universal unique ID of the service.
Service Name	Name of the service.
Service Namespace	Name of the Namespace in which the service resides.
Application	Name of the service application.
Service Type	Type of the service.
Service Protocol	Name of the service protocol.
Host IP Address	IP address of the service host.
Service Target Port	Name of the port that connects with the service.
Created Time	Timestamp at which the service was created.
Deployment Details
Deployment UUID	Universal unique ID of the deployment.
Deployment Name	Name of the deployment.
Deployment Namespace	Namespace where the deployment exists.
Deployment Replicas	The number of replicas in a deployment.
Deployment Available Replicas	Number of available replicas in a deployment.
Deployment Availability	Availability of the deployment.

Persistent Volumes

Parameter	Description	Data Collection Method
		Kubectl	Azure Monitor
PERSISTENT VOLUMES (PV) DETAILS
PV UUID	Universal unique ID of the Persistent Volume.
PV Name	Name of the Persistent Volume.
PV Status	Status of the Persistent Volume. Possible values: Available Bound Released Failed Pending
PV Claim	Name of the Persistent Volume Claim.
PV Access Mode	The mode through which you can access the Persistent Volume.
PV Storage Class	Name of the Persistent Volume storage class.
PV Capacity	The capacity of the Persistent Volume (in GiB).
PV Created Time	Timestamp at which the Persistent Volume was created.
PERSISTENT VOLUMES CLAIM (PVC) DETAILS
PVC UUID	Universal unique ID of the persistent volume.
PVC Name	Name of the Persistent Volume Claim.
PVC Namespace	Name of the Namespace in which the Claim exists.
PVC Status	Status of the Persistent Volume Claim. Possible values: Available Bound Released Failed Pending
PV Name	Name of the Persistent Volume.
PVC Access Mode	The mode through which you can access the Persistent Volume Claim.
PVC Storage Class	Name of the Persistent Volume storage class.
PVC Requests	Number of Persistent Volume Claim requests (in GiB).
PVC Created Time	Timestamp at which the Persistent Volume Claim was created.

Service Map

• Displays a graphical map view containing namespace and service details.
• All the namespace with its status and pods count for each phase will be seen inside cluster circle.
• Green color indicates that the namespace is UP and red color indicates it is DOWN.
• The cluster services under a namespace can be seen branching as a tree.
• Each service contains its host IP address and port details.