IManagedClusterAgentPoolProfile Interface

The list of Availability zones to use for nodes. This can only be specified if the AgentPoolType property is 'VirtualMachineScaleSets'.

AKS will associate the specified agent pool with the Capacity Reservation Group.

Number of agents (VMs) to host docker containers. Allowed values must be in the range of 0 to 1000 (inclusive) for user pools and in the range of 1 to 1000 (inclusive) for system pools. The default value is 1.

This is the ARM ID of the source object to be used to create the target object.

Whether to enable auto-scaler

Whether to enable host based OS and data drive encryption. This is only supported on certain VM sizes and in certain Azure regions. For more information, see: https://docs.microsoft.com/azure/aks/enable-host-encryption

Whether to use a FIPS-enabled OS. See Add a FIPS-enabled node pool for more details.

Whether each node is allocated its own public IP. Some scenarios may require nodes in a node pool to receive their own dedicated public IP addresses. A common scenario is for gaming workloads, where a console needs to make a direct connection to a cloud virtual machine to minimize hops. For more information see assigning a public IP per node. The default is false.

Whether to enable UltraSSD

Unique read-only string used to implement optimistic concurrency. The eTag value will change when the resource is updated. Specify an if-match or if-none-match header with the eTag value for a subsequent request to enable optimistic concurrency per the normal etag convention.

The Gateway agent pool associates one public IPPrefix for each static egress gateway to provide public egress. The size of Public IPPrefix should be selected by the user. Each node in the agent pool is assigned with one IP from the IPPrefix. The IPPrefix size thus serves as a cap on the size of the Gateway agent pool. Due to Azure public IPPrefix size limitation, the valid value range is [28, 31] (/31 = 2 nodes/IPs, /30 = 4 nodes/IPs, /29 = 8 nodes/IPs, /28 = 16 nodes/IPs). The default value is 31.

GPUInstanceProfile to be used to specify GPU MIG instance profile for supported GPU VM SKU.

Whether to install GPU drivers. When it's not specified, default is Install.

The fully qualified resource ID of the Dedicated Host Group to provision virtual machines from, used only in creation scenario and not allowed to changed once set. This is of the form: /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.Compute/hostGroups/{hostGroupName}. For more information see Azure dedicated hosts.

Allowed list of unsafe sysctls or unsafe sysctl patterns (ending in *).

The maximum number of container log files that can be present for a container. The number must be ≥ 2.

The maximum size (e.g. 10Mi) of container log file before it is rotated.

If CPU CFS quota enforcement is enabled for containers that specify CPU limits. The default is true.

The CPU CFS quota period value. The default is '100ms.' Valid values are a sequence of decimal numbers with an optional fraction and a unit suffix. For example: '300ms', '2h45m'. Supported units are 'ns', 'us', 'ms', 's', 'm', and 'h'.

The CPU Manager policy to use. The default is 'none'. See Kubernetes CPU management policies for more information. Allowed values are 'none' and 'static'.

If set to true it will make the Kubelet fail to start if swap is enabled on the node.

The percent of disk usage after which image garbage collection is always run. To disable image garbage collection, set to 100. The default is 85%

The percent of disk usage before which image garbage collection is never run. This cannot be set higher than imageGcHighThreshold. The default is 80%

The maximum number of processes per pod.

The Topology Manager policy to use. For more information see Kubernetes Topology Manager. The default is 'none'. Allowed values are 'none', 'best-effort', 'restricted', and 'single-numa-node'.

Determines the placement of emptyDir volumes, container runtime data root, and Kubelet ephemeral storage.

The size in MB of a swap file that will be created on each node.

Sysctl settings for Linux agent nodes.

Whether the kernel should make aggressive use of memory compaction to make more hugepages available. Valid values are 'always', 'defer', 'defer+madvise', 'madvise' and 'never'. The default is 'madvise'. For more information see Transparent Hugepages.

Whether transparent hugepages are enabled. Valid values are 'always', 'madvise', and 'never'. The default is 'always'. For more information see Transparent Hugepages.

The maximum number of nodes for auto-scaling

The maximum number of pods that can run on a node.

Message of the day for Linux nodes, base64-encoded. A base64-encoded string which will be written to /etc/motd after decoding. This allows customization of the message of the day for Linux nodes. It must not be specified for Windows nodes. It must be a static string (i.e., will be printed raw and not be executed as a script).

The minimum number of nodes for auto-scaling

The mode of an agent pool. A cluster must have at least one 'System' Agent Pool at all times. For additional information on agent pool restrictions and best practices, see: https://docs.microsoft.com/azure/aks/use-system-pools

Unique name of the agent pool profile in the context of the subscription and resource group. Windows agent pool names must be 6 characters or less.

The port ranges that are allowed to access. The specified ranges are allowed to overlap.

The IDs of the application security groups which agent pool will associate when created.

IPTags of instance-level public IPs.

The version of node image

The node labels to be persisted across all nodes in agent pool.

The public IP prefix ID which VM nodes should use IPs from. This is of the form: /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.Network/publicIPPrefixes/{publicIPPrefixName}

The taints added to new nodes during node pool create and scale. For example, key=value:NoSchedule.

OS Disk Size in GB to be used to specify the disk size for every machine in the master/agent pool. If you specify 0, it will apply the default osDisk size according to the vmSize specified.

The OS disk type to be used for machines in the agent pool. The default is 'Ephemeral' if the VM supports it and has a cache disk larger than the requested OSDiskSizeGB. Otherwise, defaults to 'Managed'. May not be changed after creation. For more information see Ephemeral OS.

The operating system type. The default is Linux.

Pod IP Allocation Mode. The IP allocation mode for pods in the agent pool. Must be used with podSubnetId. The default is 'DynamicIndividual'.

The ID of the subnet which pods will join when launched. If omitted, pod IPs are statically assigned on the node subnet (see vnetSubnetID for more details). This is of the form: /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.Network/virtualNetworks/{virtualNetworkName}/subnets/{subnetName}

Tells whether the cluster is Running or Stopped

The error additional info.

The error code.

The error details.

The error message.

The error target.

The current deployment or provisioning state.

The ID for Proximity Placement Group.

The scale down mode to use when scaling the Agent Pool. This also effects the cluster autoscaler behavior. If not specified, it defaults to Delete.

Specifications on how to scale the VirtualMachines agent pool to a fixed size.

The Virtual Machine Scale Set eviction policy to use. This cannot be specified unless the scaleSetPriority is 'Spot'. If not specified, the default is 'Delete'.

The Virtual Machine Scale Set priority. If not specified, the default is 'Regular'.

Secure Boot is a feature of Trusted Launch which ensures that only signed operating systems and drivers can boot. For more details, see aka.ms/aks/trustedlaunch. If not specified, the default is false.

vTPM is a Trusted Launch feature for configuring a dedicated secure vault for keys and measurements held locally on the node. For more details, see aka.ms/aks/trustedlaunch. If not specified, the default is false.

SSH access method of an agent pool.

The max price (in US Dollars) you are willing to pay for spot instances. Possible values are any decimal value greater than zero or -1 which indicates default price to be up-to on-demand. Possible values are any decimal value greater than zero or -1 which indicates the willingness to pay any on-demand price. For more details on spot pricing, see spot VMs pricing

The tags to be persisted on the agent pool virtual machine scale set.

The type of Agent Pool.

The drain timeout for a node. The amount of time (in minutes) to wait on eviction of pods and graceful termination per node. This eviction wait time honors waiting on pod disruption budgets. If this time is exceeded, the upgrade fails. If not specified, the default is 30 minutes.

The maximum number or percentage of nodes that are surged during upgrade. This can either be set to an integer (e.g. '5') or a percentage (e.g. '50%'). If a percentage is specified, it is the percentage of the total agent pool size at the time of the upgrade. For percentages, fractional nodes are rounded up. If not specified, the default is 10%. For more information, including best practices, see: https://learn.microsoft.com/en-us/azure/aks/upgrade-cluster

The maximum number or percentage of nodes that can be simultaneously unavailable during upgrade. This can either be set to an integer (e.g. '1') or a percentage (e.g. '5%'). If a percentage is specified, it is the percentage of the total agent pool size at the time of the upgrade. For percentages, fractional nodes are rounded up. If not specified, the default is 0. For more information, including best practices, see: https://learn.microsoft.com/en-us/azure/aks/upgrade-cluster

The soak duration for a node. The amount of time (in minutes) to wait after draining a node and before reimaging it and moving on to next node. If not specified, the default is 0 minutes.

Defines the behavior for undrainable nodes during upgrade. The most common cause of undrainable nodes is Pod Disruption Budgets (PDBs), but other issues, such as pod termination grace period is exceeding the remaining per-node drain timeout or pod is still being in a running state, can also cause undrainable nodes.

The status of nodes in a VirtualMachines agent pool.

The size of the agent pool VMs. VM size availability varies by region. If a node contains insufficient compute resources (memory, cpu, etc) pods might fail to run correctly. For more details on restricted VM sizes, see: https://docs.microsoft.com/azure/aks/quotas-skus-regions

The ID of the subnet which agent pool nodes and optionally pods will join on startup. If this is not specified, a VNET and subnet will be generated and used. If no podSubnetID is specified, this applies to nodes and pods, otherwise it applies to just nodes. This is of the form: /subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.Network/virtualNetworks/{virtualNetworkName}/subnets/{subnetName}

Whether to disable OutboundNAT in windows nodes. The default value is false. Outbound NAT can only be disabled if the cluster outboundType is NAT Gateway and the Windows agent pool does not have node public IP enabled.

Determines the type of workload a node can run.