Prerequisites for use

Baidu AI Cloud’s Cloud Container Engine (CCE) offers standard managed clusters where core components and container-related technical architectures are comprehensively managed. For unmanaged components and applications running in CCE clusters, improper operations might lead to failures. To ensure reliable service stability and minimize risks, thoroughly review the suggestions and precautions in this document before using the CCE container service.

Operation and maintenance of control plane components

• The Master, Etcd, and other managed components (such as the CCE LB Controller) in CCE managed clusters are maintained by the Baidu AI Cloud technical team. CCE ensures high cluster availability and provides both automatic and manual scaling as needed. Users only need to purchase the worker nodes necessary for running their workloads in the cluster.
• The Master and Etcd components are not categorized as user resources. Therefore, you cannot independently modify the scaling or service parameters of these components.

Note: For CCE standalone clusters, users are responsible for managing and maintaining the Master and Etcd components.

Operation and maintenance of data plane components

The data plane components of CCE managed clusters, such as CoreDNS, CCE Network Plugin, Kube-Proxy, and Kubelet, operate on user nodes (e.g., BCC servers). Therefore, ensuring stable operation and availability requires collaborative maintenance by both CCE and users.

CCE provides the following support for data plane components:

• Provides features such as parameterized configuration, functionality optimization, bug fixes, and CVE patches for components, accompanied by detailed guides.
• Delivers observable capabilities such as monitoring, logging, and alerting. Core components generate logs that can be integrated with Baidu AI Cloud Log Service (BLS) for collection and analysis.
• For other issues encountered during use, CCE generally offers relevant technical guidance and support.

When using data plane components, please follow the following suggestions:

• Ensure you use the latest component versions, as these are regularly updated to address bugs and introduce new features. Once CCE releases a new component version, choose a suitable time to upgrade, prioritizing service stability. Follow the instructions in the component upgrade guide for the upgrade process.
• Some data plane components are influenced by upstream open-source community versions and may encounter bugs or vulnerabilities. Please ensure timely upgrades to prevent service disruptions caused by these issues in open-source components.
• Avoid directly using the OpenAPIs of IaaS products to modify the runtime environment of cluster components, such as changing the node running status, security group configurations, network settings, or BLB configurations linked to the cluster via load balancer OpenAPIs. Unauthorized alterations to IaaS resources may result in issues with data plane components.
• When managing data plane components, please configure custom parameters of components through the Cloud Container Engine Console Cluster Management Page Component Management module or via OpenAPI. Modifying component configurations through other channels may cause component function abnormalities, and parameters may be automatically overwritten in some cases.
• It is recommended to enable container observability in the cluster and configure alerts for the corresponding basic resources and service resources. CCE will send important cluster control plane and data plane metrics to the Cloud Managed Service for Prometheus. For more information, please refer to Monitor Clusters with Prometheus.