The autonomous intelligent O&M feature for Container Service Agent enables proactive operations by automatically detecting cluster exceptions and triggering the agent to analyze and resolve them.
Overview
The autonomous intelligent O&M feature is powered by the ack-aiops-guard component. It uses autonomous exception awareness rules to monitor observable signals in the cluster, such as changes in Kubernetes resource status, abnormal events, and external monitoring alerts. A rule engine then analyzes these signals and triggers the Container Service Agent to begin an O&M process.
The following video demonstrates the entire process, from enabling the feature to the agent autonomously resolving a cluster exception.
Enable and authorize the feature
This feature is currently in a phased rollout. To use it, please submit a ticket.
To enable the Container Service Agent feature, you must complete the Container Service Agent authorization. The autonomous intelligent O&M feature is an optional add-on for the Container Service Agent, which requires you to additionally authorize the AliyunCSManagedAiopsGuardRole with permissions for the autonomous anomaly detection component.
On the ACK Cluster List page, click the name of your target cluster. In the navigation pane on the left of the cluster details page, select ACK Agent.
On the Tasks tab, if the Container Service Agent chat capability is not enabled, click Authorize Now to authorize the service.
Switch to the Autonomous Intelligent O&M tab. Follow the on-screen instructions to complete Role Authorization (granting the
AliyunCSManagedAiopsGuardRole) and Component Installation (installing the ack-aiops-guard component). Completing these steps enables the autonomous intelligent O&M feature for your cluster.
Autonomous exception awareness rules
New versions of the ack-aiops-guard component include updated autonomous exception awareness rules. Upgrade to the latest version to get the newest rules.
June 2026
v1.1.1
Name | Monitored object | Description |
CrashLoopBackOff | Pod | A Pod enters the CrashLoopBackOff state. |
ImagePullBackOff | Pod | An image pull fails. |
PodOOMKill | Pod | A Pod is OOMKilled. |
ContainerOOMKilled | container | A container is OOMKilled. |
PodEvicted | Pod | A Pod is evicted. |
NodeStatusChange | node | A node's status condition changes. |
JobFailure | Job | A Job fails. |
K8sWarningEvent | event | A Warning-type event occurs. |