Overview of Lingjun node pools

Updated at:

In an ACK Lingjun managed cluster, a Lingjun node pool groups Lingjun compute nodes so that container runtime settings, CPU policies, labels, taints, and user data apply to every node in the pool. Node pools are also the unit for scheduling workloads to specific nodes or GPU configurations.

Lingjun node pools cannot be created from the ACK console. To create one, log on to the Intelligent Computing Lingjun console, create a cluster, and bind a group to it. For setup instructions, see Create an ACK Lingjun managed cluster.

How it works

Intelligent Computing Lingjun organizes compute nodes into groups. Assign the compute nodes in a cluster to different groups based on your environment, cluster, and node requirements to improve node utilization. A Lingjun node pool maps one Lingjun group to one ACK node pool, bringing Lingjun nodes under standard Kubernetes node pool workflows.

Each Lingjun node belongs to exactly one Lingjun node pool.

Features

Lingjun node pools support the same features as ACK node pools. Each node in the pool shares the following configurable attributes:

Attribute

Description

Container runtime

Runtime type and version used by all nodes in the pool

CPU management policy

CPU allocation policy used by workloads on the nodes

Labels and taints

Kubernetes labels and taints used to control scheduling

User data

User data script run after the node joins the cluster

image

Default label and taint

Every node in a Lingjun node pool is automatically assigned the following label and taint, which prevent ACK system components from scheduling pods to Lingjun nodes:

Type

Key

Value or effect

Label

alibabacloud.com/lingjun-worker

true

Taint

node-role.alibabacloud.com/lingjunEffect

NoSchedule

To run a pod on a Lingjun node, add a toleration for the taint to the pod spec.

Important

Do not delete the default label or taint. If removed, ACK system components may schedule pods to Lingjun nodes, which can affect your workloads.

What's next