After you enable the auto upgrade feature for ACK Serverless cluster, you can keep your cluster upgraded on a schedule and reduce the O&M workload of version management. After you configure an auto upgrade policy, the system generates a cluster upgrade plan in advance, then runs a precheck and upgrades the cluster control plane within the maintenance window, so you can focus on developing container applications.
Limits
Auto upgrade is supported only for ACK Serverless cluster running Kubernetes 1.22 or later.
Why use auto upgrade
For outdated cluster versions, ACK Serverless cluster no longer release new features or fix functional and security defects, and provide only limited technical support. For more information, see Risks of Expired Versions. Running a supported version better ensures the security and stability of your cluster.
Cluster auto upgrade reduces the O&M workload of cluster version management. After you enable auto upgrade, you can periodically upgrade your cluster to a supported new version and use the features that the new version provides.
Auto upgrade policy
Based on the auto upgrade policy that you configure and the version release schedule for ACK Serverless cluster, cluster auto upgrade generates an upgrade plan for a specific cluster and runs the upgrade at the time scheduled in the plan. The auto upgrade policy supports configuring the cluster maintenance window and the upgrade frequency. Unlike a manual upgrade, enabling cluster auto upgrade does not upgrade the cluster immediately. If you need to complete a cluster upgrade quickly, manually upgrade the cluster. For instructions, see Manually upgrade a cluster.
During an actual upgrade, auto upgrade and manual upgrade follow the same process and provide the same capabilities. The two are not mutually exclusive. That is, after you enable auto upgrade, you can still perform a manual upgrade.
Like a manual upgrade, an auto upgrade does not support rollback. For more upgrade considerations, see Manually upgrade a cluster.
Cluster maintenance window
The cluster maintenance window applies at the cluster level and takes effect for the auto upgrade of cluster versions. Cluster auto upgrade generates an upgrade plan based on the cluster maintenance window that you configure, and runs the precheck and the upgrade only within the maintenance window that you specify.
If you modify the cluster maintenance window before an upgrade plan runs, the scheduled plan is automatically canceled. If you modify the maintenance window while a cluster auto upgrade is in progress, the cluster upgrade that has already started is canceled.
To allow enough time for the upgrade, the cluster maintenance window must be at least two hours long.
Upgrade frequency
You can select different upgrade frequencies to meet the needs of different use cases.
For example, your cluster currently runs 1.26.3-aliyun.1. Assume that Container Service for Kubernetes (ACK) clusters support three upgradeable versions: 1.26.15-aliyun.1, 1.28.9-aliyun.1, and 1.30.1-aliyun.1. Of these, 1.30 is the latest released version and 1.28 is the second-latest released version. The following table describes each upgrade frequency and the specific version that your cluster is automatically upgraded to under each frequency.
Upgrade frequency | Description | Target upgrade version |
Latest patch version (Patch) | When a patch version is available for the current minor version, the cluster is automatically upgraded to that patch version. The new Kubernetes version does not contain breaking changes. | 1.26.15-aliyun.1. |
Second-latest minor version (Stable) | The cluster is automatically upgraded to the latest patch version of the second-latest minor version. The new Kubernetes version may include changes to APIs, functions, and features, but its stability has been widely verified. | 1.28.9-aliyun.1. |
Latest minor version (Rapid) | The cluster is automatically upgraded to the latest patch version of the latest minor version, so you can get new features from the Kubernetes community faster. | 1.30.1-aliyun.1. |
For more information about patch versions and minor versions, see Version Guide.
Upgrade plan
Based on your auto upgrade policy and the cluster version release schedule of Container Service for Kubernetes (ACK), the system checks whether your cluster meets the conditions for auto upgrade. If your cluster meets the conditions, the system generates an auto upgrade plan at least one day in advance and provides the upgrade plan information in the console and through OpenAPI. If you need to skip this upgrade plan under special circumstances, cancel it manually. Canceling an upgrade plan applies only to that single plan and does not disable auto upgrade. ACK Serverless cluster still deliver auto upgrade jobs within the cluster maintenance window.
After you configure an auto upgrade policy, the system does not run the upgrade immediately. Instead, it schedules upgrade plans in advance based on the overall scheduling policy of ACK and delivers the plans in batches after a period of time. After an upgrade plan is generated, you can view the page, or call the ListOperationPlans API to get the upgrade plan.
The precheck checks API server compatibility and add-on compatibility for your cluster. If the precheck fails, fix the issues promptly and wait for the system to run the precheck again. You can view the failure details on the Upgrade Cluster page in the History section.
Upgrade notifications
If an upgrade fails, ACK Serverless cluster send notifications through email and internal site messages.
Enable cluster auto upgrade
After you enable cluster auto upgrade, ACK Serverless cluster perform an auto upgrade within one of the cluster maintenance windows based on the overall scheduling of ACK.
-
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, enable auto upgrade, and see the preceding Auto upgrade policy to configure the cluster upgrade policy and the maintenance window.
Before the auto upgrade job runs, the system automatically triggers a precheck. If the precheck fails, fix the issues as prompted on the page. You can view the precheck and upgrade status and the job details in the History section.
If you need to cancel this upgrade plan under special circumstances, cancel it manually as soon as possible.
(Optional) After the upgrade is complete, view the cluster version on the Clusters page to confirm that the upgrade is successful.
References
When auto upgrade cannot meet your requirements, you can also choose a manual upgrade:
Console: Manually upgrade a cluster
OpenAPI: UpgradeCluster
If you run large-scale workloads in a production environment but are using ACK Serverless Basic cluster, a hot migration to ACK Serverless Pro clusters is recommended. For instructions, see Hot migrate a Basic cluster to a Pro cluster.
For the support policy and lifecycle of Kubernetes versions in ACK Serverless cluster, see Version Guide. You can also view the description of a specific version to learn about its upgrade considerations, new features, deprecated functions, and deprecated APIs.