To mitigate security and stability risks associated with outdated versions, promptly upgrade your cluster. Upgrade the control plane and the node pools so that your entire cluster runs the new version.
Before upgrading the cluster, review Upgrade a cluster to understand the upgrade process, methods, and important considerations.
Operation entry point
First, upgrade the control plane. Then, upgrade the node pools. For clusters that run version 1.30 or later, the highest supported control plane version can be two minor versions higher than the lowest kubelet version among all nodes. For example, if the lowest kubelet version among all nodes is 1.30, the control plane can be upgraded to 1.32 at most.
-
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 .
On the Upgrade Cluster page, select an available Destination Version and follow the on-screen instructions to complete the upgrade process.
Upgrade the control plane
1. Pre-check
The control plane upgrade pre-check scans for items such as deprecated APIs, and checks component compatibility and the overall cluster status.
For clusters that run version 1.20 or later, the pre-check scans for any deprecated APIs. The scan results do not affect the upgrade process. They are informational only. Address any issues before upgrading to ensure that the new cluster version operates as expected.
On the Control Plane Update tab, click Precheck to scan for potential risks prior to the upgrade. After the scan completes, view the results in the Pre-check Results section. The following figure provides an example:
Normal: The upgrade check is successful. Proceed with the upgrade.
Abnormal: The result does not affect the operation or status of the current cluster. Address the issues based on the provided solutions. For more information, see Cluster check items and solutions.
2. Perform the upgrade
Duration: ACK managed clusters and ACK serverless clusters, the upgrade is managed by ACK and takes approximately 5 minutes. For ACK dedicated clusters, the master nodes are upgraded one by one in sequence. Each node takes approximately 8 minutes.
After resolving the pre-check issues, click Upgrade and follow the on-screen instructions to upgrade the control plane. During the upgrade, do not deploy or modify business applications, and do not perform O&M operations such as cluster scaling or node pool management. Such operations may interrupt the upgrade or cause service disruptions.
After the control plane is upgraded, nodes added by scale-out also run the control plane version.
3. Post-upgrade validation
After the control plane is upgraded, check the following items:
The control plane has been upgraded. The cluster version is updated on the Clusters page.
The API Server and core components are in a Normal state.
Business applications operate as expected.
Pods can be created normally.
Nodes can be added normally.
4. (Optional) Continue to upgrade the control plane
To move the control plane to a later version, repeat steps 1 to 3 to run the control plane pre-check, perform the upgrade, and validate the result. Repeat this cycle until the control plane reaches the highest version available for the cluster.
Upgrade a node pool
After the control plane is upgraded, upgrade the node pool as soon as possible during off-peak hours. The node pool upgrade involves upgrading the kubelet and container runtime on the nodes.
1. Pre-check
The node pool upgrade pre-check scans the node status, system resources, disk status, and the network environment.
On the Node Pool Upgrade page, locate the destination node pool in the list and click Upgrade in the Actions column. Then, at the bottom of the page, click Precheck to scan for potential risks during the upgrade process. After the scan completes, view the results in the Pre-check Results section.
Normal: The upgrade check is successful. You can proceed with the upgrade.
Abnormal: The result does not affect the operation or status of the current cluster. Address the issues based on the provided solutions. For more information, see Cluster check items and solutions.
2. Configure an upgrade policy and perform the upgrade
Duration: The duration depends on how the nodes are batched. An in-place upgrade takes approximately 5 to 10 minutes per batch. A disk replacement upgrade without snapshots takes approximately 8 minutes. The specific duration is influenced by the node draining time. To create snapshots, the upgrade starts after the snapshots are created. The time required to create snapshots depends on the data volume.
On the Update a node pool tab, find the destination node pool in the node pool list and click Update Now.
In the Upgrade Policy Configuration panel, confirm the upgrade information, configure the upgrade policy as described in the following table, and then click Start Update to complete the node pool upgrade as prompted.
Configuration item | Description |
Version information | The current and available versions of kubelet and the container runtime. |
Update Nodes | Upgrade All Nodes, or select Specific Node and upgrade only those nodes. |
Update Method |
|
Ignore Warnings | Specifies whether to continue the node pool upgrade when the pre-check that runs automatically returns warning items. |
Batch Update Policy |
|
3. Post-upgrade validation
After the node pool is upgraded, check the following items:
The nodes have been upgraded. The kubelet and containerd versions are updated on the node details page.
Pod scheduling is functioning normally.
Business applications operate as expected.
References
For answers to common questions before and during a cluster upgrade, see FAQ.
For information about ACK's Kubernetes version support policy, see Version Guide.
As of version 1.24, Docker is no longer supported as a built-in container runtime. If you upgrade to version 1.24 or later, you must migrate the container runtime of nodes from Docker to containerd.
As of version 1.30, CentOS and Alibaba Cloud Linux 2 are no longer supported. Use other supported operating systems.
Alternatively, enable automatic cluster upgrades to reduce the O&M workload associated with version management. For more information, see Automatically upgrade a cluster.
Before you upgrade, add any orphan nodes in the cluster to node pools so that all nodes are managed in a unified manner, as described in Migrate unmanaged nodes to a node pool.