The Container Service AI Assistant (AI Assistant), an intelligent feature from the Alibaba Cloud Container Service team, uses large language models. The AI Assistant combines expertise in Kubernetes and cloud-native fields, operational observable data, and the diagnostic experience of ACK technical experts to provide large language model-enhanced intelligent diagnostics and AIOps capabilities for ACK and Kubernetes.
Usage Notice
Welcome to the AI Assistant! Before using it, read the following notice carefully:
This service generates content using large language models. Due to technical limitations and the nature of large language models, Alibaba Cloud cannot guarantee the completeness and accuracy of the generated content. Exercise caution when evaluating the generated content. Do not over-rely on it. If your judgments or actions based on the generated content cause any loss, you bear full responsibility and consequences.
Ensure your input information is legal and compliant. Do not harm public order, infringe on others' legitimate rights, or fabricate or spread false information. You are responsible for your usage.
During service use, the system collects status data related to resources under your account (such as ECS instances, ACK clusters) for problem solving and abnormal diagnostics. Alibaba Cloud does not collect your user business data.
Core Features
Full Lifecycle O&M Management
The AI Assistant is designed to make Alibaba Cloud Container Service easier to use and learn. It simplifies O&M for container service clusters and Kubernetes resource objects by providing intelligent assistance during the following stages:
Day-0 Design and Planning Stage: Receive assistance with cloud-native containerization planning and design. For example, the AI Assistant can recommend configurations for resources such as clusters and nodes.
Day-1 Resource and Application Creation Stage: Receive assistance with designing and writing application configuration files, such as Terraform and Kubernetes YAML files.
Day-2 O&M Stage: Receive assistance with intelligent and rapid problem diagnostics in AIOps, recommendations for cost and resource usage policies, and optimizations for high availability, stability, and security solutions.
AI-assisted Intelligent Container Service Observability Features
The AI Assistant deeply integrates powerful language model capabilities with ACK's observability best practices. You can use natural language to easily query and analyze various observable data from clusters, which elevates the troubleshooting experience from "data retrieval" to "intelligent insight."
Prometheus: Supports metric queries for Alibaba Cloud Prometheus and self-built Prometheus instances associated with your ACK clusters. It also converts natural language into PromQL.
Cluster Control Plane Log Query: Supports querying control plane SLS logs for ACK clusters. You can run SLS SQL queries or convert natural language to SLS-SQL.
Audit Log: Tracks Kubernetes operation audits.
Container Service Intelligent Diagnostics and Health Check
On the ACK console, you can view abnormal node status, abnormal pods, error messages for abnormal workloads, and events. Click the Intelligent Diagnostics button to launch the AI Assistant. It quickly diagnoses the status of abnormal clusters and applications.
The AI Assistant combines observable data, such as abnormal statuses and events, from nodes and applications in the current cluster. It performs a comprehensive analysis and provides intelligent diagnostics. Typically, the AI Assistant explains the cause of the abnormal status and suggests solutions. If the AI Assistant cannot pinpoint the root cause or provide a usable solution, you can initiate further diagnostics to perform in-depth troubleshooting for abnormal pods and nodes based on the diagnostic experience of ACK technical experts. For more information, see Use cluster diagnostics.
Agent Feature Support List
Prerequisites
Activate Container Service for Kubernetes and complete the product authorization. For more information, see Quickly create an ACK managed cluster.
Authorize the AI Assistant Agent feature for Container Service.
Alibaba Cloud account: No additional authorization is required.
RAM user: If you are a RAM user and are using the AI Assistant for the first time, you must add an access policy that includes the
cs:CreateSessionMessageandcs:UpdateMessageFeedbackpermissions. For more information, see Manage RAM role permissions.
Enable the AI Assistant in the console
The AI Assistant provides features such as AI chat and intelligent diagnostics. To use AI chat, you can launch the AI Assistant from any page in the console. You can also launch the AI Assistant through intelligent diagnostics when cluster resources, such as nodes, pods, workloads, or events, are abnormal. This lets you quickly diagnose the status of abnormal clusters and applications.
Direct Invocation
On any page of the console, you can click the floating AI Assistant icon. Then, follow the on-screen prompts to start a conversation and use the AI chat feature.
Launch via intelligent diagnostics
Go to a resource page in the console, such as Workloads. Locate the abnormal resource. Then, click Intelligent Diagnostics in the status bar.
The AI Assistant then provides intelligent answers related to the abnormal issue to help you locate and troubleshoot the problem.
Manage historical session memory
Each time you launch the ACK AI chat or intelligent diagnostics, the AI Assistant starts a new session. Recent conversations are recorded in the session history, which the AI Assistant uses as context for its responses. To start a new conversation, you can refresh the page or click the button to clear the message history.
Contact Us
If you have any comments or suggestions while using the AI Assistant, join the DingTalk group (DingTalk group ID: 70080006301) to contact us.