Super nodes are created in TKE Serverless clusters for Pod scheduling. A super node is created in each subnet of the container network when you create a TKE Serverless cluster.
In the following scenarios, you can resolve the issues by creating super nodes:
When you run a large-scale workload in a TKE Serverless cluster, a Pod cannot be created due to exhaustion of IPs in the subnet of availability zone where the service is located. It is recommended that you create super nodes in a new subnet to have more IPs.
As the scale of services expands, services need to be automatically distributed in multiple availability zones. It is recommended that you create super nodes in a new subnet to expand availability zones of the resources.
When you create a workload, Pod pending is caused by resource shortage for the availability zone. This may be reflected as the following cluster events:
EVENT REASON : “FailedCreatePodSandBox”
EVENT MESSAGE : “Failed to create pod sandbox in underlay (will retry): insufficient resource”
You can create super nodes in other availability zones to expand resources available in the cluster.
Fees are not charged for super nodes and the cost will be calculated based on the CPU, GPU, memory value and the running time of the workload. For details, see Billing Overview, Product Pricing, and Purchase Limits.
This document describes how to create a super node in a TKE Serverless cluster in the TKE console.
After the creation of the super node, please refer to Managing a Super Node for subsequent management.
Was this page helpful?