日日操夜夜添-日日操影院-日日草夜夜操-日日干干-精品一区二区三区波多野结衣-精品一区二区三区高清免费不卡

公告：魔扣目錄網為廣大站長提供免費收錄網站服務，提交前請做好本站友鏈：【 網站目錄：http://www.ylptlb.cn 】， 免友鏈快審服務（50元/站），

點擊這里在線咨詢客服
新站提交
  • 網站：51998
  • 待審：31
  • 小程序：12
  • 文章：1030137
  • 會員：747

pod污點taint?與容忍度tolerations詳解

發布時間：2024-03-09 00:02:23 作者：網友整理

目錄
  • 一.系統環境
  • 二.前言
  • 三.污點taint
    • 3.1 污點taint概覽
    • 3.2 給節點添加污點taint
  • 四.容忍度tolerations
    • 4.1 容忍度tolerations概覽
    • 4.2 設置容忍度tolerations

一.系統環境

服務器版本 docker軟件版本 Kubernetes(k8s)集群版本 CPU架構
CentOS Linux release 7.4.1708 (Core) Docker version 20.10.12 v1.21.9 x86_64

Kubernetes集群架構：k8scloude1作為master節點，k8scloude2，k8scloude3作為worker節點

服務器 操作系統版本 CPU架構 進程 功能描述
k8scloude1/192.168.110.130 CentOS Linux release 7.4.1708 (Core) x86_64 docker，kube-apiserver，etcd，kube-scheduler，kube-controller-manager，kubelet，kube-proxy，coredns，calico k8s master節點
k8scloude2/192.168.110.129 CentOS Linux release 7.4.1708 (Core) x86_64 docker，kubelet，kube-proxy，calico k8s worker節點
k8scloude3/192.168.110.128 CentOS Linux release 7.4.1708 (Core) x86_64 docker，kubelet，kube-proxy，calico k8s worker節點

二.前言

本文介紹污點taint 與容忍度tolerations，可以影響pod的調度。

使用污點taint 與容忍度tolerations的前提是已經有一套可以正常運行的Kubernetes集群，關于Kubernetes(k8s)集群的安裝部署，可以查看博客《Centos7 安裝部署Kubernetes(k8s)集群》

三.污點taint

3.1 污點taint概覽

節點親和性 是 Pod 的一種屬性，它使 Pod 被吸引到一類特定的節點 （這可能出于一種偏好，也可能是硬性要求）。 污點（Taint） 則相反——它使節點能夠排斥一類特定的 Pod。

3.2 給節點添加污點taint

給節點增加一個污點的語法如下：給節點 node1 增加一個污點，它的鍵名是 key1，鍵值是 value1，效果是 NoSchedule。 這表示只有擁有和這個污點相匹配的容忍度的 Pod 才能夠被分配到 node1 這個節點。

#污點的格式：鍵=值:NoSchedule
kubectl taint nodes node1 key1=value1:NoSchedule
#只有鍵沒有值的話，格式為：鍵:NoSchedule
kubectl taint nodes node1 key1:NoSchedule

移除污點語法如下：

kubectl taint nodes node1 key1=value1:NoSchedule-

節點的描述信息里有一個Taints字段，Taints字段表示節點有沒有污點

[root@k8scloude1 deploy]# kubectl get nodes -o wide
NAME         STATUS   ROLES                  AGE   VERSION   INTERNAL-IP       EXTERNAL-IP   OS-IMAGE                KERNEL-VERSION          CONTAINER-RUNTIME
k8scloude1   Ready    control-plane,master   8d    v1.21.0   192.168.110.130   <none>        CentOS Linux 7 (Core)   3.10.0-693.el7.x86_64   docker://20.10.12
k8scloude2   Ready    <none>                 8d    v1.21.0   192.168.110.129   <none>        CentOS Linux 7 (Core)   3.10.0-693.el7.x86_64   docker://20.10.12
k8scloude3   Ready    <none>                 8d    v1.21.0   192.168.110.128   <none>        CentOS Linux 7 (Core)   3.10.0-693.el7.x86_64   docker://20.10.12
[root@k8scloude1 deploy]# kubectl describe nodes k8scloude1
Name:               k8scloude1
Roles:              control-plane,master
Labels:             beta.kubernetes.io/arch=amd64
                    beta.kubernetes.io/os=linux
                    kubernetes.io/arch=amd64
                    kubernetes.io/hostname=k8scloude1
                    kubernetes.io/os=linux
                    node-role.kubernetes.io/control-plane=
                    node-role.kubernetes.io/master=
                    node.kubernetes.io/exclude-from-external-load-balancers=
Annotations:        kubeadm.alpha.kubernetes.io/cri-socket: /var/run/dockershim.sock
                    node.alpha.kubernetes.io/ttl: 0
                    projectcalico.org/IPv4Address: 192.168.110.130/24
                    projectcalico.org/IPv4IPIPTunnelAddr: 10.244.158.64
                    volumes.kubernetes.io/controller-managed-attach-detach: true
CreationTimestamp:  Sun, 09 Jan 2022 16:19:06 +0800
Taints:             node-role.kubernetes.io/master:NoSchedule
Unschedulable:      false
......

查看節點是否有污點，Taints: node-role.kubernetes.io/master:NoSchedule表示k8s集群的master節點有污點，這是默認就存在的污點，這也是master節點為什么不能運行應用pod的原因。

[root@k8scloude1 deploy]# kubectl describe nodes k8scloude2 | grep -i Taints
Taints:             <none>
[root@k8scloude1 deploy]# kubectl describe nodes k8scloude1 | grep -i Taints
Taints:             node-role.kubernetes.io/master:NoSchedule
[root@k8scloude1 deploy]# kubectl describe nodes k8scloude3 | grep -i Taints
Taints:             <none>

創建pod，nodeSelector:kubernetes.io/hostname: k8scloude1表示pod運行在標簽為kubernetes.io/hostname=k8scloude1的節點上。

關于pod的調度詳細內容，請查看博客《pod(八)：pod的調度——將 Pod 指派給節點》

[root@k8scloude1 pod]# vim schedulepod4.yaml 
[root@k8scloude1 pod]# cat schedulepod4.yaml
apiVersion: v1
kind: Pod
metadata:
  creationTimestamp: null
  labels:
    run: pod1
  name: pod1
  namespace: pod
spec:
  nodeSelector:
    kubernetes.io/hostname: k8scloude1
  containers:
  - image: nginx
    imagePullPolicy: IfNotPresent
    name: pod1
    resources: {}
    ports:
    - name: http
      containerPort: 80
      protocol: TCP
      hostPort: 80
  dnsPolicy: ClusterFirst
  restartPolicy: Always
status: {}

標簽為kubernetes.io/hostname=k8scloude1的節點為k8scloude1節點

[root@k8scloude1 pod]# kubectl get nodes -l kubernetes.io/hostname=k8scloude1
NAME         STATUS   ROLES                  AGE   VERSION
k8scloude1   Ready    control-plane,master   8d    v1.21.0

創建pod，因為k8scloude1上有污點，pod1不能運行在k8scloude1上，所以pod1狀態為Pending

[root@k8scloude1 pod]# kubectl apply -f schedulepod4.yaml 
pod/pod1 created
 #因為k8scloude1上有污點，pod1不能運行在k8scloude1上，所以pod1狀態為Pending
[root@k8scloude1 pod]# kubectl get pod -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP       NODE     NOMINATED NODE   READINESS GATES
pod1   0/1     Pending   0          9s    <none>   <none>   <none>           <none>
[root@k8scloude1 pod]# kubectl delete pod pod1 --force
warning: Immediate deletion does not wait for confirmation that the running resource has been terminated. The resource may continue to run on the cluster indefinitely.
pod "pod1" force deleted
[root@k8scloude1 pod]# kubectl get pod -o wide
No resources found in pod namespace.

四.容忍度tolerations

4.1 容忍度tolerations概覽

容忍度（Toleration） 是應用于 Pod 上的。容忍度允許調度器調度帶有對應污點的 Pod。 容忍度允許調度但并不保證調度：作為其功能的一部分， 調度器也會評估其他參數。

污點和容忍度（Toleration）相互配合，可以用來避免 Pod 被分配到不合適的節點上。 每個節點上都可以應用一個或多個污點，這表示對于那些不能容忍這些污點的 Pod， 是不會被該節點接受的。

4.2 設置容忍度tolerations

只有擁有和這個污點相匹配的容忍度的 Pod 才能夠被分配到 node節點。

查看k8scloude1節點的污點

[root@k8scloude1 pod]# kubectl describe nodes k8scloude1 | grep -i taint
Taints:             node-role.kubernetes.io/master:NoSchedule

你可以在 Pod 規約中為 Pod 設置容忍度，創建pod，tolerations參數表示可以容忍污點：node-role.kubernetes.io/master:NoSchedule ，nodeSelector:kubernetes.io/hostname: k8scloude1表示pod運行在標簽為kubernetes.io/hostname=k8scloude1的節點上。

[root@k8scloude1 pod]# vim schedulepod4.yaml 
[root@k8scloude1 pod]# cat schedulepod4.yaml 
apiVersion: v1
kind: Pod
metadata:
  creationTimestamp: null
  labels:
    run: pod1
  name: pod1
  namespace: pod
spec:
  tolerations:
  - key: "node-role.kubernetes.io/master"
    operator: "Equal"
    value: ""
    effect: "NoSchedule"
  nodeSelector:
    kubernetes.io/hostname: k8scloude1
  containers:
  - image: nginx
    imagePullPolicy: IfNotPresent
    name: pod1
    resources: {}
    ports:
    - name: http
      containerPort: 80
      protocol: TCP
      hostPort: 80
  dnsPolicy: ClusterFirst
  restartPolicy: Always
status: {}
[root@k8scloude1 pod]# kubectl get pods -o wide
No resources found in pod namespace.
[root@k8scloude1 pod]# kubectl apply -f schedulepod4.yaml 
pod/pod1 created

查看pod，即使k8scloude1節點有污點，pod還是正常運行。

taint污點和cordon，drain的區別：某個節點上有污點，可以設置tolerations容忍度，讓pod運行在該節點，某個節點被cordon，drain，則該節點不能被分配出去運行pod。

關于cordon，drain的詳細信息，請查看博客《cordon節點，drain驅逐節點，delete 節點》

[root@k8scloude1 pod]# kubectl get pods -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP              NODE         NOMINATED NODE   READINESS GATES
pod1   1/1     Running   0          4s    10.244.158.84   k8scloude1   <none>           <none>
[root@k8scloude1 pod]# kubectl delete pod pod1 --force
warning: Immediate deletion does not wait for confirmation that the running resource has been terminated. The resource may continue to run on the cluster indefinitely.
pod "pod1" force deleted
[root@k8scloude1 pod]# kubectl get pods -o wide
No resources found in pod namespace.

注意，tolerations容忍度有兩種寫法，任選一種即可:

tolerations:
- key: "key1"
  operator: "Equal"
  value: "value1"
  effect: "NoSchedule"
tolerations:
- key: "key1"
  operator: "Exists"
  effect: "NoSchedule"

給k8scloude2節點打標簽

[root@k8scloude1 pod]# kubectl label nodes k8scloude2 taint=T
node/k8scloude2 labeled
[root@k8scloude1 pod]# kubectl get node --show-labels
NAME         STATUS   ROLES                  AGE   VERSION   LABELS
k8scloude1   Ready    control-plane,master   8d    v1.21.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8scloude1,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=,node.kubernetes.io/exclude-from-external-load-balancers=
k8scloude2   Ready    <none>                 8d    v1.21.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8scloude2,kubernetes.io/os=linux,taint=T
k8scloude3   Ready    <none>                 8d    v1.21.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8scloude3,kubernetes.io/os=linux

對k8scloude2設置污點

#污點taint的格式：鍵=值:NoSchedule
[root@k8scloude1 pod]# kubectl taint node k8scloude2 wudian=true:NoSchedule
node/k8scloude2 tainted
[root@k8scloude1 pod]# kubectl describe nodes k8scloude2 | grep -i Taints
Taints:             wudian=true:NoSchedule

創建pod，tolerations參數表示容忍污點wudian=true:NoSchedule，nodeSelector:taint: T參數表示pod運行在標簽為nodeSelector=taint: T的節點。

[root@k8scloude1 pod]# vim schedulepod4.yaml 
[root@k8scloude1 pod]# cat schedulepod4.yaml 
apiVersion: v1
kind: Pod
metadata:
  creationTimestamp: null
  labels:
    run: pod1
  name: pod1
  namespace: pod
spec:
  tolerations:
  - key: "wudian"
    operator: "Equal"
    value: "true"
    effect: "NoSchedule"
  nodeSelector:
    taint: T
  containers:
  - image: nginx
    imagePullPolicy: IfNotPresent
    name: pod1
    resources: {}
    ports:
    - name: http
      containerPort: 80
      protocol: TCP
      hostPort: 80
  dnsPolicy: ClusterFirst
  restartPolicy: Always
status: {}
[root@k8scloude1 pod]# kubectl get pod -o wide
No resources found in pod namespace.
[root@k8scloude1 pod]# kubectl apply -f schedulepod4.yaml 
pod/pod1 created

查看pod，k8scloude2節點就算有污點也能運行pod

[root@k8scloude1 pod]# kubectl get pods -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP               NODE         NOMINATED NODE   READINESS GATES
pod1   1/1     Running   0          8s    10.244.112.177   k8scloude2   <none>           <none>
[root@k8scloude1 pod]# kubectl delete pod pod1 --force
warning: Immediate deletion does not wait for confirmation that the running resource has been terminated. The resource may continue to run on the cluster indefinitely.
pod "pod1" force deleted
[root@k8scloude1 pod]# kubectl get pods -o wide
No resources found in pod namespace.

污點容忍的另一種寫法：operator: "Exists"，沒有value值。

[root@k8scloude1 pod]# vim schedulepod4.yaml 
[root@k8scloude1 pod]# cat schedulepod4.yaml 
apiVersion: v1
kind: Pod
metadata:
  creationTimestamp: null
  labels:
    run: pod1
  name: pod1
  namespace: pod
spec:
  tolerations:
  - key: "wudian"
    operator: "Exists"
    effect: "NoSchedule"
  nodeSelector:
    taint: T
  containers:
  - image: nginx
    imagePullPolicy: IfNotPresent
    name: pod1
    resources: {}
    ports:
    - name: http
      containerPort: 80
      protocol: TCP
      hostPort: 80
  dnsPolicy: ClusterFirst
  restartPolicy: Always
status: {}
[root@k8scloude1 pod]# kubectl apply -f schedulepod4.yaml 
pod/pod1 created

查看pod，k8scloude2節點就算有污點也能運行pod

[root@k8scloude1 pod]# kubectl get pods -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP               NODE         NOMINATED NODE   READINESS GATES
pod1   1/1     Running   0          10s   10.244.112.178   k8scloude2   <none>           <none>
[root@k8scloude1 pod]# kubectl delete pod pod1 --force
warning: Immediate deletion does not wait for confirmation that the running resource has been terminated. The resource may continue to run on the cluster indefinitely.
pod "pod1" force deleted
[root@k8scloude1 pod]# kubectl get pods -o wide
No resources found in pod namespace.

給k8scloude2節點再添加一個污點

[root@k8scloude1 pod]# kubectl describe nodes k8scloude2 | grep Taints
Taints:             wudian=true:NoSchedule
[root@k8scloude1 pod]# kubectl taint node k8scloude2 zang=shide:NoSchedule
node/k8scloude2 tainted
[root@k8scloude1 pod]# kubectl describe nodes k8scloude2 | grep Taints
Taints:             wudian=true:NoSchedule
[root@k8scloude1 pod]# kubectl describe nodes k8scloude2 | grep -A2 Taints
Taints:             wudian=true:NoSchedule
                    zang=shide:NoSchedule
Unschedulable:      false
[root@k8scloude1 pod]# kubectl describe nodes k8scloude2 | grep -A1 Taints
Taints:             wudian=true:NoSchedule
                    zang=shide:NoSchedule

創建pod，tolerations參數表示容忍2個污點：wudian=true:NoSchedule和zang=shide:NoSchedule，nodeSelector:taint: T參數表示pod運行在標簽為nodeSelector=taint: T的節點。

[root@k8scloude1 pod]# vim schedulepod4.yaml 
[root@k8scloude1 pod]# cat schedulepod4.yaml 
apiVersion: v1
kind: Pod
metadata:
  creationTimestamp: null
  labels:
    run: pod1
  name: pod1
  namespace: pod
spec:
  tolerations:
  - key: "wudian"
    operator: "Equal"
    value: "true"
    effect: "NoSchedule"
  - key: "zang"
    operator: "Equal"
    value: "shide"
    effect: "NoSchedule"
  nodeSelector:
    taint: T
  containers:
  - image: nginx
    imagePullPolicy: IfNotPresent
    name: pod1
    resources: {}
    ports:
    - name: http
      containerPort: 80
      protocol: TCP
      hostPort: 80
  dnsPolicy: ClusterFirst
  restartPolicy: Always
status: {}
[root@k8scloude1 pod]# kubectl apply -f schedulepod4.yaml 
pod/pod1 created

查看pod，k8scloude2節點就算有2個污點也能運行pod

[root@k8scloude1 pod]# kubectl get pods -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP               NODE         NOMINATED NODE   READINESS GATES
pod1   1/1     Running   0          6s    10.244.112.179   k8scloude2   <none>           <none>
[root@k8scloude1 pod]# kubectl delete pod pod1 --force
warning: Immediate deletion does not wait for confirmation that the running resource has been terminated. The resource may continue to run on the cluster indefinitely.
pod "pod1" force deleted

創建pod，tolerations參數表示容忍污點：wudian=true:NoSchedule，nodeSelector:taint: T參數表示pod運行在標簽為nodeSelector=taint: T的節點。

[root@k8scloude1 pod]# vim schedulepod4.yaml 
[root@k8scloude1 pod]# cat schedulepod4.yaml 
apiVersion: v1
kind: Pod
metadata:
  creationTimestamp: null
  labels:
    run: pod1
  name: pod1
  namespace: pod
spec:
  tolerations:
  - key: "wudian"
    operator: "Equal"
    value: "true"
    effect: "NoSchedule"
  nodeSelector:
    taint: T
  containers:
  - image: nginx
    imagePullPolicy: IfNotPresent
    name: pod1
    resources: {}
    ports:
    - name: http
      containerPort: 80
      protocol: TCP
      hostPort: 80
  dnsPolicy: ClusterFirst
  restartPolicy: Always
status: {}
[root@k8scloude1 pod]# kubectl apply -f schedulepod4.yaml 
pod/pod1 created

查看pod，一個節點有兩個污點值，但是yaml文件只容忍一個，所以pod創建不成功。

[root@k8scloude1 pod]# kubectl get pods -o wide
NAME   READY   STATUS    RESTARTS   AGE   IP       NODE     NOMINATED NODE   READINESS GATES
pod1   0/1     Pending   0          8s    <none>   <none>   <none>           <none>
[root@k8scloude1 pod]# kubectl delete pod pod1 --force
warning: Immediate deletion does not wait for confirmation that the running resource has been terminated. The resource may continue to run on the cluster indefinitely.
pod "pod1" force deleted
[root@k8scloude1 pod]# kubectl get pods -o wide
No resources found in pod namespace.

取消k8scloude2污點

[root@k8scloude1 pod]# kubectl describe nodes k8scloude2 | grep -A2 Taints
Taints:             wudian=true:NoSchedule
                    zang=shide:NoSchedule
Unschedulable:      false
#取消污點
[root@k8scloude1 pod]# kubectl taint node k8scloude2 zang-
node/k8scloude2 untainted
[root@k8scloude1 pod]# kubectl taint node k8scloude2 wudian-
node/k8scloude2 untainted
[root@k8scloude1 pod]# kubectl describe nodes k8scloude1 | grep -A2 Taints
Taints:             node-role.kubernetes.io/master:NoSchedule
Unschedulable:      false
Lease:
[root@k8scloude1 pod]# kubectl describe nodes k8scloude2 | grep -A2 Taints
Taints:             <none>
Unschedulable:      false
Lease:
[root@k8scloude1 pod]# kubectl describe nodes k8scloude3 | grep -A2 Taints
Taints:             <none>
Unschedulable:      false
Lease:

Tips:如果自身機器有限，只能有一臺機器，則可以把master節點的污點taint取消，就可以在master上運行pod了。

以上就是pod污點taint 與容忍度tolerations詳解的詳細內容，更多關于污點taint容忍度tolerations 的資料請關注其它相關文章！

分享到：
標簽：Pod 容忍度 服務器 污點 詳解
用戶無頭像

網友整理

注冊時間：

網站：5 個   小程序：0 個  文章：12 篇

  • 51998

    網站

  • 12

    小程序

  • 1030137

    文章

  • 747

    會員

趕快注冊賬號，推廣您的網站吧！
最新入駐小程序

數獨大挑戰2018-06-03

數獨一種數學游戲，玩家需要根據9

答題星2018-06-03

您可以通過答題星輕松地創建試卷

全階人生考試2018-06-03

各種考試題，題庫，初中，高中，大學四六

運動步數有氧達人2018-06-03

記錄運動步數，積累氧氣值。還可偷

每日養生app2018-06-03

每日養生,天天健康

體育訓練成績評定2018-06-03

通用課目體育訓練成績評定