Container Service for Kubernetes:Enable node auto scaling

Before you begin

To make the most of node auto scaling, read Node scaling and understand the following concepts:

How node auto scaling works and its features

Use cases where node auto scaling can meet your business requirements

Important considerations before using node auto scaling

During a scale-in, subscription instances are removed but not released. To avoid extra costs, use pay-as-you-go instances when you enable this feature.

Prerequisites

• Make sure you have activated Auto Scaling.

• See Usage notes to understand the quotas and limitations of node scaling.

• Node auto scaling has known limitations with certain scheduling policies, which may lead to unexpected scaling results. If your workloads or components use an unsupported scheduling policy, consider one of the following solutions:

  • Solution 1: Switch to node instant scaling.

  • Solution 2: Deploy the affected workloads or components to a node pool without node scaling enabled.

    For example, to deploy the ack-node-local-dns-admission-controller component, place it in a node pool without node scaling enabled and declare the following node affinity requirement in the component's configuration:

    nodeAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
        nodeSelectorTerms:
        - matchExpressions:
          - key: "k8s.aliyun.com"
            operator: "NotIn"
            values: ["true"]

• The cluster-autoscaler component requires node resources for updates or deployments. Insufficient resources may cause these operations to fail and lead to scaling issues. Ensure that your nodes have adequate resources.

This feature involves the following steps:

1. Step 1: Enable node auto scaling for the cluster: You must first enable node auto scaling at the cluster level before the scaling policies of your node pools can take effect.

2. Step 2: Configure a node pool for auto scaling: The node auto scaling feature only affects node pools that are configured for auto scaling. Therefore, you must set the scaling mode of the target node pools to Auto.

Step 1: Enable node auto scaling

1. Log on to the ACK console. In the left navigation pane, click Clusters.

2. On the Clusters page, click the name of your cluster. In the left navigation pane, click Nodes > Node Pools.

3. On the Node Pools page, click Enable next to Node Scaling.

4. If you are using node auto scaling for the first time, follow the on-screen instructions to activate the Auto Scaling service and grant the required permissions. You can skip this step if you have already done this.

   • ACK managed cluster: Authorize the AliyunCSManagedAutoScalerRole role.

   • ACK dedicated cluster: Authorize the KubernetesWorkerRole role and attach the AliyunCSManagedAutoScalerRolePolicy.

     In the Node Scaling Configuration dialog box, after the precheck passes, click the RAM role link (such as KubernetesWorkerRole-xxxx) to complete authorization in the RAM console.

5. In the Node Scaling Configuration dialog box, set Node Scaling Plan to Auto Scaling, configure the scaling parameters, and then click OK.

   You can switch the node scaling method after the initial configuration. To do this, change the selection here to node instant scaling. Carefully read and follow the on-screen instructions to complete the process.

   Parameter	Description
   Node Pool Scale-out Policy	Random Policy: If multiple node pools are available for scale-out, one is chosen at random. Default Policy: If multiple node pools are available for scale-out, the one that results in the least resource waste is chosen. Priority-based Policy: If multiple node pools are available for scale-out, the one with the highest priority is chosen. Node pool priority is defined by the Node Pool Scale-out Priority parameter.
   Node Pool Scale-out Priority	Sets the scale-out priority for node pools. This parameter takes effect only when Node Pool Scale-out Policy is set to Priority-based Policy. The value can be an integer from 1 to 100. A larger value indicates a higher priority. Click Add next to the parameter, select a node pool with auto scaling enabled, and set its priority. If no node pools with auto scaling enabled are available, you can ignore this parameter for now and configure it after you complete Step 2: Configure a node pool for auto scaling.
   Scaling Sensitivity	The interval at which the system checks for scaling conditions. The default value is 60s.
   During auto scaling, the scaling component automatically triggers scale-out based on scheduling conditions. Important ECS nodes: A node scale-in can occur only when all three conditions are met: Scale-in Threshold, Scale-in Trigger Delay, and Cooldown Period. GPU nodes: A GPU node scale-in can occur only when all three conditions are met: GPU Scale-in Threshold, Scale-in Trigger Delay, and Cooldown Period.
   Allow Scale-in	Specifies whether to allow node scale-in operations. If disabled, scale-in settings do not take effect. Use this setting with caution.
   Scale-in Threshold	The ratio of total resource requests to the total capacity of a single node in a node pool where node auto scaling is enabled. A node is considered for scale-in only when this ratio is below the configured threshold, meaning both its CPU and memory utilization are below the Scale-in Threshold.
   GPU Scale-in Threshold	The scale-in threshold for GPU instances. A GPU node is considered for scale-in only when its CPU, memory, and GPU utilization are all below the GPU Scale-in Threshold.
   Scale-in Trigger Delay	The time to wait from when a scale-in condition is detected to when the scale-in operation is actually performed. Unit: minutes. Default value: 10 minutes. Important The scaling component performs a node scale-in only after the Scale-in Threshold is met and the Scale-in Trigger Delay period has passed.
   Cooldown Period	The period after a scale-out event during which the scaling component will not perform a scale-in. During the cooldown period, the scaler does not perform scale-ins but continues to evaluate nodes against the scale-in conditions. After the cooldown ends, if a node has met the scale-in threshold for longer than the scale-in delay, the scaler removes it. For example, if the cooldown is 10 minutes and the scale-in delay is 5 minutes, the scaler will not scale in any nodes for 10 minutes after the last scale-out. However, during these 10 minutes, it still checks for nodes that are eligible for scale-in. Once the 10-minute cooldown ends, if a node has met the scale-in threshold for more than the 5-minute delay, it is scaled in.

   View advanced settings

   Parameter	Description
   Pod Termination Timeout	Maximum wait time for pod termination during scale-in. Unit: seconds. If a pod is not evicted before the timeout, the node is not released.
   Minimum Pod Replicas	Scale-in protection threshold. Nodes with ReplicationController or ReplicaSet pods are not scaled in if the replica count falls below this value. Applies only to ReplicationController and ReplicaSet pods, not StatefulSet or DaemonSet.
   Enable DaemonSet Pod Eviction	When enabled, DaemonSet pods are evicted when their node is scaled in.
   Skip nodes with pods in the kube-system namespace	When enabled, nodes with kube-system namespace pods are excluded from scale-in. Note This does not apply to DaemonSet pods or mirror pods.

Step 2: Configure a node pool

You can either configure an existing node pool by changing its Scaling Mode to Auto, or create a new node pool with auto scaling enabled.

For more information, see Create and manage node pools. The key parameters are described below:

Step 3: (Optional) Verify the results

After you complete the steps, node auto scaling is active. The node pool's status will indicate that auto scaling has started, and the cluster-autoscaler component will be automatically installed.

FAQ