[英]kubectl top nodes command gives only CPU and Memory usage, how to get GPU usage information in Kubernetes cluster?
I'm using Kubernetes 1.7 and I have a GPU machine in the cluster. 我正在使用Kubernetes 1.7,并且集群中有一台GPU机器。 The kubectl top nodes
gives only CPU and Memory usage like this kubectl top nodes
仅提供这样的CPU和内存使用率
NAME CPU(cores) CPU% MEMORY(bytes) MEMORY%
machine1 183m 1% 5284Mi 8%
machine2 181m 1% 5543Mi 8%
machine3 250m 1% 7682Mi 7%
machine4 328m 5% 1578Mi 16%
Is there a way to get GPU usage ? 有没有办法获得GPU使用率 ?
I'm afraid that kubectl top
does not provide GPU metrics, regardless if drivers are installed or not. 恐怕无论是否安装驱动程序, kubectl top
都不提供GPU指标。
To inspect GPU statistics, I recommend using Heapster monitoring tool and Prometheus that provides the extension to get the accelerator metrics. 要检查GPU统计信息,我建议使用Heapster监视工具和Prometheus ,后者提供了获取加速器指标的扩展。
At last, you may be interested in Nvidia GPU works for Kubernetes and GCE concepts of using this kind of hardware. 最后,您可能会对Nvidia GPU用于Kubernetes的作品以及使用这种硬件的GCE概念感兴趣。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.