Auto Mode provides automated and intelligent cluster O&M capabilities for enterprise production environments. By enabling Auto Mode, you can create a Kubernetes cluster that adheres to best practices with simplified network configuration, and extend its capabilities with automated cluster management.
With Auto Mode, you can benefit from:
Comprehensive managed O&M: Fully managed control plane and key system components, automatic Kubernetes version upgrades, and full lifecycle management for nodes.
Automatic node scaling: Eliminates the need for manual capacity planning by automatically scaling nodes up or down, optimizing your resource costs.
Default security best practices: Enhances system security with an immutable node file system.
Auto Mode is suitable for general containerized workloads, especially when:
You cannot dedicate significant human resources to Kubernetes infrastructure management.
You lack professional Kubernetes O&M experience but want to use Kubernetes to containerize and modernize applications.
You need to quickly deploy standardized Kubernetes environments across multiple regions.
Key features
Managed O&M:
Cluster-level management: A fully managed control plane and key system components (such as DNS servers and monitoring agents), with automatic Kubernetes version upgrades enabled by default.
Node-level management: A node pool with Auto Mode enabled is created by default, with Container Service for Kubernetes (ACK) automatically recommending optimal instance types. ACK manages all node-level O&M tasks, including OS and component upgrades and security patching.
Elastic node scaling: Auto Mode includes built-in node instant scaling, which dynamically adds or removes nodes based on workload demands. This eliminates the need for upfront capacity planning, and optimizes resource costs.
Enhanced security:
Optimized software stack: The default ContainerOS uses an immutable root file system to enhance security. Its streamlined system configuration also accelerates node startup times and optimizes hardware performance.
Default security and observability: Features such as cluster inspection, API server auditing, and control plane component log collection are enabled by default to help you quickly locate and resolve cluster issues.
Workload deployment
Auto Mode is designed based on the concepts of elastic capacity, immutable infrastructure, and zero O&M. It is important to identify potential compatibility issues for workloads that depend on node-level customizations or local persistent storage. Note the following when you deploy workloads in a node pool with Auto Mode enabled:
Avoid manual node operations: ACK will perform automated O&M on nodes, including OS upgrades, software patching, and reboots. To avoid conflicts with automated policies, do not perform manual operations on the Elastic Compute Service (ECS) nodes in the pool, such as rebooting, attaching data disks, or modifying configurations by logging onto the node.
Configure workloads for disruption: Ensure your workloads are configured to gracefully handle pod evictions during node O&M. Set appropriate numbers of replicas and configure PreStop hooks and PodDisruptionBudgets to ensure your services remain available.
Use persistent storage: The underlying OS, ContainerOS, uses an immutable root file system. Since nodes can be scaled down and terminated at any time, avoid using node system path storage, such as HostPath. Instead, use PersistentVolumeClaims (PVCs) for persistent storage.
Get started with Auto Mode
Deploy a workload in the cluster and expose it using an ALB Ingress. For more information, see Deploy workloads and configure load balancing.
Create and manage additional node pools (with or without Auto Mode) as needed.
Usage notes
Shared responsibility model
While Auto Mode is designed to automate and simplify Kubernetes O&M, you are still responsible for certain aspects of cluster management.
Alibaba Cloud's responsibilities | Your responsibilities | Shared responsibilities |
|
|
|
Billing
Auto Mode is a feature of ACK managed Pro clusters and does not incur additional charges. You will be billed for the cluster management fees and any associated cloud resources. For details, see Billing overview and Cloud resource fees.
Quotas and limitations
By default, an Auto Mode node pool can scale up to a maximum of 200 nodes. You can adjust this limit in the node pool settings.
Auto Mode node pools do not support instance types such as Arm, GPU, or local disk instances. The default instance types recommended by ACK meet the requirements of most application scenarios. Adjust the instance types in the console as needed to improve the resilience of your node pool and prevent scaling failures.
For large-scale clusters, be aware of the quotas and limits for ACK and its dependent cloud services, including the following:
ACK configurations, such as account balance.
Single-cluster capacity limits that specify the maximum capacity of different Kubernetes resources in a single cluster.
Quotas for dependent cloud services, such as ECS and VPC, and how to increase them.
FAQ
Can I create a non-Auto Mode node pool in an Auto Mode cluster?
Yes. You can create self-managed node pools alongside Auto Mode ones. Doing so allows you to flexibly choose the level of management for different workloads. For feature comparison, see Comparison of managed node pool configurations