Supported operating system images for ACK - Container Service for Kubernetes

Supported operating system images and usage limits for Container Service for Kubernetes nodes.

Supported operating system images for Container Service for Kubernetes

When creating an ACK cluster or node pool, select an operating system image for the nodes. Only the latest version of each image type is available by default.

Important

ACK maintains its own supported image list, which may lag behind the latest ECS versions. This list is authoritative.
Alibaba Cloud Linux 2 and CentOS reached end of life (EOL) on March 31, 2024 and June 30, 2024, respectively. Alibaba Cloud no longer provides technical support for ECS instances running these versions. Use Alibaba Cloud Linux 3 or ContainerOS instead. See [Product Change] End of Maintenance for Alibaba Cloud Linux 2 and CentOS 7.

Image type	Image name (including version)	Image ID	Platform	System architecture	Boot mode	System size	Supported cluster versions
Alibaba Cloud Linux 4 LTS 64-bit Container-Optimized Edition	Alibaba Cloud Linux 4 LTS 64-bit Container-Optimized Edition	aliyun_4_x64_20G_container_optimized_alibase_20251106.vhd Uses cgroup v2 by default.	Aliyun	x86_64	UEFI-Preferred	20 GiB	1.33 and later
Alibaba Cloud Linux 3.2104 LTS 64-bit Container-Optimized Edition	Alibaba Cloud Linux 3.2104 Container-Optimized	aliyun_3_x64_20G_container_optimized_alibase_20251215.vhd Uses cgroup v2 by default.	Aliyun	x86_64	UEFI-Preferred	20 GiB	1.26 and later
ContainerOS	ContainerOS 3	lifsea_3_x64_5G_alibase_20251204.qcow2 ContainerOS 3.3 and later use cgroup v2 by default.	Aliyun	x86_64	BIOS	5 GiB	1.31 and later
Alibaba Cloud Linux 3	Alibaba Cloud Linux 3.2104 LTS 64 bit	aliyun_3_x64_20G_alibase_20251215.vhd	Aliyun	x86_64	BIOS	20 GiB	1.18 and later
Alibaba Cloud Linux 3 Arm Edition	Alibaba Cloud Linux 3.2104 LTS 64 bit ARM Edition	aliyun_3_arm64_20G_alibase_20240528.vhd	Aliyun	arm64	UEFI	20 GiB	1.20 and later
Alibaba Cloud Linux UEFI 3	Alibaba Cloud Linux UEFI 3.2104 Security Enhanced	aliyun_3_x64_20G_uefi_alibase_20230727.vhd	Aliyun	x86_64	UEFI	20 GiB	1.18 and later
Red Hat	Red Hat Enterprise Linux (RHEL) 9.3 64-bit	Red Hat Enterprise Linux 9.3 64bit RHEL 9 and later use cgroup v2 by default.	Red Hat	x86_64	BIOS	20 GiB	1.20 and later
Ubuntu	Ubuntu 24.04	ubuntu_24_04_x64_20G_alibase_20251126.vhd Ubuntu 22 and later use cgroup v2 by default.	Ubuntu	x86_64	BIOS	20 GiB	1.30 and later
Windows	Windows Server 2022 (20240220)	win2022_21H2_x64_dtc_en-us_40G_container_alibase_20240220.vhd	WindowsServer2022	x86_64	BIOS	40 GiB	1.18 and later
Windows	Windows Server 2019 (20240220)	win2019_1809_x64_dtc_en-us_40G_container_alibase_20240220.vhd	WindowsServer2019	x86_64	BIOS	40 GiB	1.18 and later
Windows Core	Windows Server Core, version 2022 (20240223)	wincore_2022_x64_dtc_en-us_40G_container_alibase_20240223.vhd	WindowsServer2022	x86_64	BIOS	40 GiB	1.18 and later
Alibaba Cloud Linux 2 (EOL)	Alibaba Cloud Linux 2.1903 LTS 64 bit	aliyun_2_1903_x64_20G_alibase_20231221.vhd	Aliyun	x86_64	BIOS	20 GiB	Earlier than 1.30
CentOS (EOL)	CentOS 7.9 64 bit	centos_7_9_x64_20G_alibase_20230718.vhd	CentOS	x86_64	BIOS	20 GiB	Earlier than 1.30

Usage notes

Operating system limits

Available operating systems depend on the selected instance types. Multiple instance types narrow the list to those supported by all.

Operating system type

Limits

Alibaba Cloud Linux 4 LTS 64-bit Container-Optimized Edition

Requires Terway version 1.16.4 or later, or a version between 1.9.18 (inclusive) and 1.10.0 (exclusive).

Alibaba Cloud Linux

Only Alibaba Cloud Linux is supported in the following scenarios:

The container runtime is Sandboxed-Container.
When you create a Confidential Computing ACK managed cluster.

Alibaba Cloud Linux 3 Arm Edition

Only Arm architecture instance types are supported.

Ubuntu

Cluster version 1.30 or later required. To upgrade, see Manually upgrade a cluster.
Automatic OS upgrades are disabled during node initialization.
During initialization, /etc/resolv.conf is symlinked to /run/systemd/resolve/stub-resolv.conf. DNS is configured by DHCP.
Features such as CPFS persistent volumes (PVs), image acceleration plugins, and security hardening are not supported.

RHEL 9.3

RHEL requires a custom image. See Create a custom image from an instance.

Use the custom image to create a node pool. All nodes in the pool deploy with this image.

Procedure

Log on to the ACK console. In the left navigation pane, click Clusters.
On the Clusters page, click the name of your cluster. In the left navigation pane, click Nodes > Node Pools.
Click Create Node Pool. For Operating System, select Custom Image, then select the image you created.
For node pool configuration, see Create and manage node pools.

Operating system of a custom image

Create custom images only from ACK-supported operating systems.
Do not create custom images from ECS instances currently running in an ACK cluster. If needed, remove the instance from the cluster first. For details, see Remove nodes.
Predefined logic in custom images may affect node initialization, container runtime, OS upgrades, and node auto-recovery in managed node pools. Before using in production, ensure thorough testing and validation.
Do not enable a swap partition.

Windows

Limits apply to cluster type, version, network plugin, and more.

Limits

Item	Description
Cluster	An ACK managed cluster is created. You can add Windows node pools only to existing ACK managed clusters; you cannot add them during cluster creation. The cluster version is 1.32 or later. To upgrade a cluster, see Manually upgrade an ACK cluster. The cluster's network type must be Flannel. The kube-flannel-ds-windows version is v0.15.2-aliyun or later.
Container runtime	Only containerd 1.6 or later supports Windows node pools. To upgrade the runtime version, see Upgrade a node pool.
Node pool type	A Windows node pool must be a non-managed node pool.
Instance type	Support for Windows node pools varies by instance type. If node pool creation fails, the instance type you selected may be unsupported. Try again with a different one. For example, ecs.g6.xlarge supports Windows node pools. The instance type must have at least 4 vCPUs and 8 GB of memory. Windows containers are not terminated by the Out of Memory (OOM) killer when they exceed their memory limits. Since May 2021, for ACK clusters of v1.16 or later, newly added Windows nodes reserve a portion of resources at startup (1.5 vCPUs, 2.5 GB of RAM, and 3 GB of disk space) to ensure the stability of the Windows operating system, kubelet, and container runtime. For more information about resource reservation, see Node resource reservation policy. This resource reservation helps prevent Windows nodes from becoming unavailable due to workload overallocation. However, a memory leak in a Windows container application can still cause the node to crash. Windows containers have a footprint. For more information, see Memory requirements for Windows containers.

See Create and manage Windows node pools.

cgroup versions

The Linux kernel uses cgroups (v1 and v2) to limit and isolate process resources. Migrating from v1 to v2 requires changing the node pool's operating system and adjusting workloads due to significant differences. See Migrate nodes to cgroup v2.

References

See Alibaba Cloud Linux 3 container-optimized image.
Use node pools to manage nodes for operations such as maintenance, OS CVE patching, and auto scaling.
When you create an ACK managed cluster, you only create worker nodes. ACK hosts the control plane, reducing maintenance overhead.