kubectl top no内存使用率的计算

1.从下图可以看出node:10.81.163.21的内存使用率93%，该节点总共有7821M(Capacity Memory)物理内存，然后k8s的可分配内存(Allocatable Memory)为5399696Ki

那这个98%是怎么来的?

[root@iZcwfshg0bs9fxZ system]# free -m
              total        used        free      shared  buff/cache   available
Mem:           7821        5176        1930           1         714        2294
Swap:             0           0           0
 
 
[root@iZcwfshg0bs9fxZ system]# kubectl top no
NAME            CPU(cores)   CPU%   MEMORY(bytes)   MEMORY%     
10.81.163.21    478m         15%    5208Mi          98%      
[root@iZcwfshg0bs9fxZ system]# kubectl describe no 10.81.163.21
Name:               10.81.163.21
Roles:              <none>
Labels:             beta.kubernetes.io/arch=amd64
                    beta.kubernetes.io/os=linux
                    ctet-sc=node
                    kubernetes.io/hostname=10.81.163.21
Annotations:        node.alpha.kubernetes.io/ttl: 0
                    volumes.kubernetes.io/controller-managed-attach-detach: true
CreationTimestamp:  Wed, 27 Mar 2019 10:25:48 +0800
Taints:             <none>
Unschedulable:      false
Conditions:
  Type             Status  LastHeartbeatTime                 LastTransitionTime                Reason                       Message
  ----             ------  -----------------                 ------------------                ------                       -------
  MemoryPressure   False   Thu, 25 Apr 2019 18:59:45 +0800   Tue, 23 Apr 2019 20:49:53 +0800   KubeletHasSufficientMemory   kubelet has sufficient memory available
  DiskPressure     False   Thu, 25 Apr 2019 18:59:45 +0800   Wed, 27 Mar 2019 10:25:52 +0800   KubeletHasNoDiskPressure     kubelet has no disk pressure
  PIDPressure      False   Thu, 25 Apr 2019 18:59:45 +0800   Wed, 27 Mar 2019 10:25:52 +0800   KubeletHasSufficientPID      kubelet has sufficient PID available
  Ready            True    Thu, 25 Apr 2019 18:59:45 +0800   Thu, 25 Apr 2019 18:57:05 +0800   KubeletReady                 kubelet is posting ready status
Addresses:
  InternalIP:  10.81.163.21
  Hostname:    10.81.163.21
Capacity:
 cpu:                4
 ephemeral-storage:  41151808Ki
 hugepages-1Gi:      0
 hugepages-2Mi:      0
 memory:             8008848Ki
 pods:               110
Allocatable:
 cpu:                3
 ephemeral-storage:  35697000443
 hugepages-1Gi:      0
 hugepages-2Mi:      0
 memory:             5399696Ki
 pods:               110
System Info:
 Machine ID:                 20190215172108590907433256076310
 System UUID:                4F846BA6-A427-4257-91CF-E496FA5BCC45
 Boot ID:                    eeb21388-45bf-429e-84a8-ff11a46833dd
 Kernel Version:             3.10.0-957.5.1.el7.x86_64
 OS Image:                   CentOS Linux 7 (Core)
 Operating System:           linux
 Architecture:               amd64
 Container Runtime Version:  docker://18.3.1
 Kubelet Version:            v1.13.1
 Kube-Proxy Version:         v1.13.1
Non-terminated Pods:         (8 in total)
  Namespace                  Name                               CPU Requests  CPU Limits  Memory Requests  Memory Limits  AGE
  ---------                  ----                               ------------  ----------  ---------------  -------------  ---
  ctet-sc                    ask-bar-service-vsv8z              0 (0%)        0 (0%)      0 (0%)           0 (0%)         5d22h
  ctet-sc                    ask-bar-web-server-h96dx           0 (0%)        0 (0%)      0 (0%)           0 (0%)         5d22h
  ctet-sc                    cs-service-vjgl5                   100m (3%)     4 (133%)    1Gi (19%)        2Gi (38%)      39m
  ctet-sc                    exam-service-g442f                 0 (0%)        0 (0%)      0 (0%)           0 (0%)         5d23h
  ctet-sc                    microclass-async-service-v7n8t     0 (0%)        0 (0%)      0 (0%)           0 (0%)         5d23h
  default                    node-server-7bnp5                  0 (0%)        0 (0%)      0 (0%)           0 (0%)         8d
  default                    system-service-5584f55df5-hkv5t    1 (33%)       6 (200%)    1Gi (19%)        6Gi (116%)     12d
  monitoring                 node-exporter-4k4p2                112m (3%)     270m (9%)   200Mi (3%)       220Mi (4%)     29d
Allocated resources:
  (Total limits may be over 100 percent, i.e., overcommitted.)
  Resource           Requests      Limits
  --------           --------      ------
  cpu                1212m (40%)   10270m (342%)
  memory             2248Mi (42%)  8412Mi (159%)
  ephemeral-storage  0 (0%)        0 (0%)

2.内存使用率=pod used total memory/Allocatable Memory=(5176*1024)/5399696=0.9816=98%

pod used total memory就是在这台机器上的所有docker容器占用内存总和；

Allocatable Memory就是kubectl describe no 查看出该节点可分配的内存=物理内存-k8s预留内存

 

同理cpu使用率=pod used total cpu/Allocatable CPU

posted @ 2020-06-02 23:36  $world  阅读(2272)  评论(0)  编辑  收藏  举报