简体   繁体   English

尽管设置了限制,但Kubernetes并没有驱逐节点

[英]Kubernetes does not evict nodes despite limit being set

I changed the default eviction policy set by kops to include the condition memory.available<1Gi . 我更改了kops设置的默认逐出策略,以包括条件memory.available<1Gi The --eviction-hard flag is now set as: --eviction-hard标志现在设置为:

memory.available<1Gi,nodefs.available<10%,nodefs.inodesFree<5%,imagefs.available<10%,imagefs.inodesFree<5%

The available memory on one node right now is at 400Mb and has been like this for quite a while. 现在,一个节点上的可用内存为400Mb,并且已经有一段时间了。 No pod eviction is happening. 没有发生容器驱逐事件。

Why isn't the kubelet evicting pods to make room? 为什么Kubelet不能逐出豆荚来腾出空间? There's plenty of room on other nodes. 其他节点上有足够的空间。

Is there an AND between eviction conditions? 驱逐条件之间是否存在AND How can I see what the kubelet sees for memory usage? 我如何查看kubelet看到的内存使用情况?

Your pods might have quality of service class set to Guaranteed, that's possible reason they are not getting evicted. 您的广告连播可能会将服务质量等级设置为“保证”,这可能是因为它们没有被驱逐。

See: https://kubernetes.io/docs/tasks/configure-pod-container/quality-service-pod/#qos-classes 参见: https : //kubernetes.io/docs/tasks/configure-pod-container/quality-service-pod/#qos-classes

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.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM