Auto Scaling

Auto Scale-out Configuration

CCE Cluster Autoscaler comprehensively checks the resource statuses of an entire cluster. When the load of a microservice is high (for example, the CPU or memory usage is too high), it will add more pods to reduce the load.

Node Capacity Expansion Conditions

• Auto scale-out when the workload cannot be scheduled: If a workload pod cannot be scheduled, the system automatically scales out the node pool with auto scaling enabled. If the pod has been configured to be affinity for a node, the system will not automatically add more nodes.

  Such auto scaling works with an HPA policy. For details, see Using HPA and CA for Auto Scaling of Workloads and Nodes.

• User-defined policy switch: specifies whether to automatically scale out a node pool based on the node scaling policies. This function is enabled by default.

Upper limit of resources to be expanded

• Total Nodes: specifies how many nodes can be present in a cluster during scale-out. If there are more nodes in the cluster than the specified value, the cluster will not add nodes. The default value is determined by how many nodes a cluster can manage at most.
• Total Cores: specifies how many cores on all nodes can be present in a cluster during scale-out. If there are more cores in the cluster than the specified value, the cluster will not add nodes. By default, the number is not limited.
• Total Memory (GiB): specifies the upper limit of the total memory of all nodes in a cluster during scale-out. If the total memory exceeds the specified value, the cluster will not add nodes. By default, the number is not limited.

When the total number of nodes, CPUs, and memory is collected, unavailable nodes in custom node pools are included but unavailable nodes in the default node pool are not included.

Scale-out Priority

You can drag and drop node pools in the list to adjust their scale-out priorities.

Auto Scale-in Configuration

CCE Cluster Autoscaler comprehensively checks the resource statuses of an entire cluster. Once it confirms that workload pods can be scheduled and run properly, it automatically obtains nodes for scale-in.

Node Scale-in Conditions

• Node Resource Condition: When the requested cluster node resources (both CPU and memory) are lower than a certain percentage (50% by default) for a period of time (10 minutes by default), a cluster scale-in is triggered.
• Node Status Condition: If a node is unavailable for a specified period of time, the node will be automatically reclaimed. The default value is 20 minutes.
• Scale-in Exception Scenarios: When a node meets the following exception scenarios, CCE will not scale in the node even if the node resources or status meets scale-in conditions:
  1. Resources on other nodes in the cluster are insufficient.
  2. Scale-in protection is enabled on the node. To enable or disable node scale-in protection, choose Nodes in the navigation pane and then click the Nodes tab. Locate the target node, choose More, and then enable or disable node scale-in protection in the Operation column.
  3. There is a pod with the non-scale label on the node.
  4. Policies such as reliability have been configured on some containers on the node.
  5. There are non-DaemonSet containers in the kube-system namespace on the node.
  6. (Optional) A container managed by a third-party pod controller is running on a node. Third-party pod controllers are for custom workloads except Kubernetes-native workloads such as Deployments and StatefulSets. Such controllers can be created using CustomResourceDefinitions.