智算集群基础设施运维通用技术要求

General Technical Requirements for Operation and Maintenance of Intelligent Computing Cluster Infrastructure V 1.0 目录 编者...........................................................................................................................................................31.引言......................................................................................................................................................42.缩略语与术语.....................................................................................................................................53.运维参考架构.....................................................................................................................................74.关键运维组件功能要求....................................................................................................................84.1.硬件管理..................................................................................................................................84.1.1.设备纳管......................................................................................................................84.1.2.配置管理....................................................................................................................114.1.3.性能管理....................................................................................................................134.1.4.告警管理....................................................................................................................154.1.5.日志管理....................................................................................................................164.2. GPU运维................................................................................................................................174.2.1. GPU配置管理............................................................................................................174.2.2. GPU驱动管理............................................................................................................174.2.3. GPU指标监控............................................................................................................184.2.4. GPU健康检查............................................................................................................184.2.5. GPU故障自愈............................................................................................................194.2.6. GPU故障诊断............................................................................................................194.2.7.性能压测....................................................................................................................194.3.智算集群运维........................................................................................................................204.3.1.平台资源监控............................................................................................................204.3.2.性能管理....................................................................................................................214.3.3.告警管理....................................................................................................................214.3.4.日志管理....................................................................................................................224.4.网络运维................................................................................................................................224.4.1.拓扑可视化................................................................................................................224.4.2.网络检测....................................................................................................................234.4.3.网络故障诊断............................................................................................................244.4.4.网络调优....................................................................................................................254.5.作业运维................................................................................................................................254.5.1.作业监控....................................................................................................................254.5.2.作业可观测................................................................................................................254.5.3.作业劣化分析............................................................................................................254.5.4.作业故障诊断............................................................................................................264.5.5.性能调优....................................................................................................................265.修订记录............................................................................................................................................27 编者 文档参编单位： 中兴通讯股份有限公司、中国信息通信研究院、阿里云计算有限公司、上海壁仞科技股份有限公司、上海天数智芯半导体股份有限公司、浪潮电子信息产业股份有限公司、上海寒武纪信息科技有限公司、昆仑芯（北京）科技股份有限公司、瀚博半导体（上海）股份有限公司（排名不分先后） 文档参编组织： 龙蜥社区智算联盟 1.引言