如何将Kubernetes节点状态从“就绪"更改为进入“未就绪"状态通过更改CPU利用率或内存利用率或磁盘压力? [英] How to Change Kubernetes Node Status from "Ready" to "NotReady" by changing CPU Utilization or memory utilization or Disk Pressure?
问题描述
我将Kubernetes集群设置为1个主节点和1个工作节点.出于测试目的，我将CPU利用率和内存利用率提高了100％，但让Node的状态没有变为"NotReady". 我正在测试压力状态..如何将MemoryPressure的状态标志更改为true或DiskPressure或PIDPressure更改为true
I have kubernetes cluster set up of 1 master and 1 worker node.For testing purpose , I increased CPU Utlization and memory utilization upto 100%, but stilling Node is not getting status "NotReady". I am testing Pressure Status .. How to change status flag of MemoryPressure to true or DiskPressure or PIDPressure to true
这是我的主节点条件:-
Here are my master node condtions :-
条件:
Type Status LastHeartbeatTime LastTransitionTime Reason Message
---- ------ ----------------- ------------------ ------ -------
NetworkUnavailable False Wed, 27 Nov 2019 14:36:29 +0000 Wed, 27 Nov 2019 14:36:29 +0000 WeaveIsUp Weave pod has set this
MemoryPressure False Thu, 28 Nov 2019 07:36:46 +0000 Fri, 22 Nov 2019 13:30:38 +0000 KubeletHasSufficientMemory kubelet has sufficient memory available
DiskPressure False Thu, 28 Nov 2019 07:36:46 +0000 Fri, 22 Nov 2019 13:30:38 +0000 KubeletHasNoDiskPressure kubelet has no disk pressure
PIDPressure False Thu, 28 Nov 2019 07:36:46 +0000 Fri, 22 Nov 2019 13:30:38 +0000 KubeletHasSufficientPID kubelet has sufficient PID available
Ready True Thu, 28 Nov 2019 07:36:46 +0000 Fri, 22 Nov 2019 13:30:48 +0000 KubeletReady kubelet is posting ready status
以下是广告连播信息:-
Here are pods info :-
Non-terminated Pods: (8 in total)
Namespace Name CPU Requests CPU Limits Memory Requests Memory Limits AGE
--------- ---- ------------ ---------- --------------- ------------- ---
kube-system coredns-5644d7b6d9-dm8v7 100m (5%) 0 (0%) 70Mi (0%) 170Mi (2%) 22d
kube-system coredns-5644d7b6d9-mz5rm 100m (5%) 0 (0%) 70Mi (0%) 170Mi (2%) 22d
kube-system etcd-ip-172-31-28-186.us-east-2.compute.internal 0 (0%) 0 (0%) 0 (0%) 0 (0%) 22d
kube-system kube-apiserver-ip-172-31-28-186.us-east-2.compute.internal 250m (12%) 0 (0%) 0 (0%) 0 (0%) 22d
kube-system kube-controller-manager-ip-172-31-28-186.us-east-2.compute.internal 200m (10%) 0 (0%) 0 (0%) 0 (0%) 22d
kube-system kube-proxy-cw8vv 0 (0%) 0 (0%) 0 (0%) 0 (0%) 22d
kube-system kube-scheduler-ip-172-31-28-186.us-east-2.compute.internal 100m (5%) 0 (0%) 0 (0%) 0 (0%) 22d
kube-system weave-net-ct9zb 20m (1%) 0 (0%) 0 (0%) 0 (0%) 22d
Allocated resources:
(Total limits may be over 100 percent, i.e., overcommitted.)
Resource Requests Limits
-------- -------- ------
cpu 770m (38%) 0 (0%)
memory 140Mi (1%) 340Mi (4%)
ephemeral-storage 0 (0%) 0 (0%)
此处为工作节点:-
Conditions:
Type Status LastHeartbeatTime LastTransitionTime Reason Message
---- ------ ----------------- ------------------ ------ -------
NetworkUnavailable False Thu, 28 Nov 2019 07:00:08 +0000 Thu, 28 Nov 2019 07:00:08 +0000 WeaveIsUp Weave pod has set this
MemoryPressure False Thu, 28 Nov 2019 07:39:03 +0000 Thu, 28 Nov 2019 07:00:00 +0000 KubeletHasSufficientMemory kubelet has sufficient memory available
DiskPressure False Thu, 28 Nov 2019 07:39:03 +0000 Thu, 28 Nov 2019 07:00:00 +0000 KubeletHasNoDiskPressure kubelet has no disk pressure
PIDPressure False Thu, 28 Nov 2019 07:39:03 +0000 Thu, 28 Nov 2019 07:00:00 +0000 KubeletHasSufficientPID kubelet has sufficient PID available
Ready True Thu, 28 Nov 2019 07:39:03 +0000 Thu, 28 Nov 2019 07:00:00 +0000 KubeletReady kubelet is posting ready status
推荐答案
有几种方法可以使节点进入NotReady状态，但不能通过Pods.当Pod开始消耗过多内存时，Kubelet只会杀死该Pod，以精确地保护节点.
There are several ways of making a node to get into NotReady status, but not through Pods. When a Pod starts to consume too much memory kubelet will just kill that pod, to precisely protect the node.
我猜您想测试当节点发生故障时会发生什么，在这种情况下，您希望将其耗尽.换句话说，要模拟节点问题，您应该执行以下操作:
I guess you want to test what happens when a node goes down, in that case you want to drain it. In other words, to simulate node issues, you should do:
kubectl drain NODE
仍然，检查kubectl drain --help以查看在什么情况下会发生什么情况.
Still, check on kubectl drain --help to see under what circumstances what happens.
编辑
实际上，我尝试访问该节点并直接在该节点上施加压力，这是20秒内发生的事情:
I tried, actually, accessing the node and running stress on the node directly, and this is what happened within 20 seconds:
root@gke-klusta-lemmy-3ce02acd-djhm:/# stress --cpu 16 --io 8 --vm 8 --vm-bytes 2G
检查节点:
$ kubectl get no -w | grep gke-klusta-lemmy-3ce02acd-djhm
gke-klusta-lemmy-3ce02acd-djhm Ready <none> 15d v1.13.11-gke.14
gke-klusta-lemmy-3ce02acd-djhm Ready <none> 15d v1.13.11-gke.14
gke-klusta-lemmy-3ce02acd-djhm NotReady <none> 15d v1.13.11-gke.14
gke-klusta-lemmy-3ce02acd-djhm NotReady <none> 15d v1.13.11-gke.14
gke-klusta-lemmy-3ce02acd-djhm NotReady <none> 15d v1.13.11-gke.14
我正在运行非常弱的节点. 1CPU @ 4GB内存
I am running pretty weak nodes. 1CPU@4GB RAM
这篇关于如何将Kubernetes节点状态从“就绪"更改为进入“未就绪"状态通过更改CPU利用率或内存利用率或磁盘压力?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
