k8s调度器
节点硬亲和度：Pod调度时必须满足规划，如不满足Pod状态将是Penging状态。
十载的公安网站建设经验，针对设计、前端、开发、售后、文案、推广等六对一服务，响应快，48小时及时工作处理。网络营销推广的优势是能够根据用户设备显示端的尺寸不同，自动调整公安建站的显示方式，使网站能够适用不同显示终端，在浏览器中调整网站的宽度，无论在任何一种浏览器上浏览网站，都能展现优雅布局与设计，从而大程度地提升浏览体验。创新互联建站从事“公安网站设计”,“公安网站推广”以来，每个客户项目都认真落实执行。
节点软亲和度：Pod调度时先按规则进行调度，如不满足规则也会找一个不匹配的节点运行。
Pod硬亲和度，软亲和度与节点硬亲和度，软亲和度相似。
1.节点硬亲和度(节点不满足规则)
[root@k8s01 yaml]# cat pod-affinity01.yaml
apiVersion: v1
kind: Pod
metadata:
name: pod-01
spec:
affinity:
nodeAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
nodeSelectorTerms:
- matchExpressions:
- {key: zone,operator: In,values: ["one"]} --节点zone(键)标签必须是one(值)
containers:
- name: pod-01
image: nginx:latest
imagePullPolicy: Never[root@k8s01 yaml]# kubectl apply -f pod-affinity01.yaml
pod/pod-01 created
[root@k8s01 yaml]# kubectl get pods --show-labels
NAME READY STATUS RESTARTS AGE LABELS
pod-01 0/1 Pending 0 103s
[root@k8s01 yaml]# kubectl describe pods pod-01
。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。。
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Warning FailedScheduling
Warning FailedScheduling
[root@k8s01 yaml]#
- matchExpressions:
- {key: zone,operator: In,values: ["one"]} --zone是键，operator：是表达式，one是值
- {key: zone,operator: In,values: ["one","two"]} --In 是包含
- {key: ssd,operator: Exists,values: []} --Exists是存在
2.节点硬亲和度(节点满足规则)
[root@k8s01 yaml]# kubectl label node k8s02 zone=one --在k8s02节点创建标签
node/k8s02 labeled
[root@k8s01 yaml]# kubectl get pods --show-labels -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
pod-01 1/1 Running 0 6m13s 10.244.1.37 k8s02
[root@k8s01 yaml]#
3.节点软亲和度
[root@k8s01 yaml]# cat pod-affinity02.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: pod-02
spec:
replicas: 3
selector:
matchLabels:
app: myapp
template:
metadata:
name: myapp-pod
labels:
app: myapp
spec:
affinity:
nodeAffinity:
preferredDuringSchedulingIgnoredDuringExecution:
- weight: 60 --60%的Pod会调度到zone标签，one值这个节点上
preference:
matchExpressions:
- {key: zone,operator: In,values: ["one"]}
containers:
- name: pod-02
image: nginx:latest
imagePullPolicy: Never[root@k8s01 yaml]# kubectl apply -f pod-affinity02.yaml
deployment.apps/pod-02 created
[root@k8s01 yaml]# kubectl get pods --show-labels -o wide --满足标签可创建，不满足也可创建
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
pod-02-77d87986b-9bzg5 1/1 Running 0 16s 10.244.1.39 k8s02
pod-02-77d87986b-dckjq 1/1 Running 0 16s 10.244.2.42 k8s03
pod-02-77d87986b-z7v47 1/1 Running 0 16s 10.244.1.38 k8s02
[root@k8s01 yaml]#
4. Pod硬亲和度(节点不满足规则)
[root@k8s01 yaml]# kubectl get pods --show-labels
NAME READY STATUS RESTARTS AGE LABELS
nginx 1/1 Running 0 4s
app=web
[root@k8s01 yaml]# cat pod-affinity03.yaml
apiVersion: v1
kind: Pod
metadata:
name: pod-01
spec:
affinity:
podAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchExpressions:
- {key: app,operator: In,values: ["web1"]} --标签值不满足
topologyKey: kubernetes.io/hostname
containers:
- name: pod-01
image: nginx:latest
imagePullPolicy: Never[root@k8s01 yaml]# kubectl apply -f pod-affinity03.yaml
pod/pod-01 created
[root@k8s01 yaml]# kubectl get pods --show-labels -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
nginx 1/1 Running 0 8m9s 10.244.1.42 k8s02
pod-01 0/1
Pending 0 28s
[root@k8s01 yaml]#
5.Pod硬亲和度 (节点满足规则)
[root@k8s01 yaml]# kubectl get pods --show-labels --查看pod标签
NAME READY STATUS RESTARTS AGE LABELS
nginx 1/1 Running 0 4s
app=web
[root@k8s01 yaml]# cat pod-affinity04.yaml
apiVersion: v1
kind: Pod
metadata:
name: pod-01
spec:
affinity:
podAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchExpressions:
- {key: app,operator: In,values: ["web"]}
topologyKey: kubernetes.io/hostname
containers:
- name: pod-01
image: nginx:latest
imagePullPolicy: Never[root@k8s01 yaml]# kubectl apply -f pod-affinity04.yaml
pod/pod-01 created
[root@k8s01 yaml]# kubectl get pods --show-labels -o wide --创建的新Pod会找app=web的标签
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
nginx 1/1 Running 0 4m14s 10.244.1.42
k8s02
pod-01 1/1
Running 0 17s 10.244.1.43
k8s02
[root@k8s01 yaml]#
6.Pod软亲和度
[root@k8s01 yaml]# kubectl get pods --show-labels -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
nginx 1/1 Running 0 14m 10.244.1.42 k8s02
[root@k8s01 yaml]# cat pod-affinity04.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: pod-02
spec:
replicas: 3
selector:
matchLabels:
app: myapp
template:
metadata:
name: myapp-pod
labels:
app: myapp
spec:
affinity:
podAffinity:
preferredDuringSchedulingIgnoredDuringExecution:
- weight: 60
podAffinityTerm:
labelSelector:
matchExpressions:
- {key: app,operator: In,values: ["web12"]}
topologyKey: zone
containers:
- name: pod-02
image: nginx:latest
imagePullPolicy: Never[root@k8s01 yaml]# kubectl get pods --show-labels -o wide --不满足Pod软亲和度也可创建
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
nginx 1/1 Running 0 16m 10.244.1.42 k8s02
pod-02-6f96fbfdf-52gbf 1/1 Running 0 11s 10.244.2.43
k8s03
pod-02-6f96fbfdf-dl5z5 1/1 Running 0 11s 10.244.1.44
k8s02
pod-02-6f96fbfdf-f8bzn 1/1 Running 0 11s 10.244.0.55
k8s01
[root@k8s01 yaml]#
7.定义污点和容忍度
NoSchedule：不能容忍此污点的新Pod对像，不能调度到当前节点，强制性。
PreferNoSchedule：新Pod尽量不调度到此节点，若无其它节点可以调度，也允许使用当前节点调度，柔和性。
NoExecute：新Pod不能调度到当前节点，如果当前节点存在Pod也会被驱逐，强制性。
[root@k8s01 yaml]# kubectl describe node k8s02 | grep -i taints --查看污点
Taints:
[root@k8s01 yaml]# kubectl taint node k8s02 node-type=production:NoSchedule --为k8s02节点打污点
node/k8s02 tainted
[root@k8s01 yaml]# kubectl describe node k8s02 | grep -i taints
Taints: node-type=production:NoSchedule
[root@k8s01 yaml]# cat pod-affinity05.yaml
apiVersion: apps/v1 kind: Deployment metadata: name: pod-02 spec: replicas: 3 selector: matchLabels: app: myapp template: metadata: name: myapp-pod labels: app: myapp spec: containers: - name: pod-02 image: nginx:latest imagePullPolicy: Never
[root@k8s01 yaml]# kubectl apply -f pod-affinity05.yaml
deployment.apps/pod-02 created
[root@k8s01 yaml]# kubectl get pods -o wide --show-labels --k8s02节点已打污点，所以不能调度Pod对像
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
pod-02-5c54dc6489-j7tsh 1/1 Running 0 29s 10.244.0.56
k8s01
pod-02-5c54dc6489-sk5dm 1/1 Running 0 29s 10.244.2.44
k8s03
pod-02-5c54dc6489-sn5wd 1/1 Running 0 29s 10.244.2.45
k8s03
[root@k8s01 yaml]
8.删除污点
[root@k8s01 yaml]# kubectl taint node k8s02 node-type:NoSchedule- --删除 NoSchedule污点
node/k8s02 untainted
[root@k8s01 yaml]# kubectl taint node k8s02 node-type=production:PreferNoSchedule --打上污点
node/k8s02 tainted
[root@k8s01 yaml]# kubectl describe node k8s02 | grep -i node-type
Taints: node-type=production:PreferNoSchedule
[root@k8s01 yaml]# kubectl taint node k8s02 node-type:PreferNoSchedule- --删除
PreferNoSchedule污点
node/k8s02 untainted
[root@k8s01 yaml]# kubectl describe node k8s02 | grep -i node-type
[root@k8s01 yaml]# kubectl taint node k8s02 node-type- --删除所有污点
node/k8s02 untainted
[root@k8s01 yaml]# kubectl describe node k8s02 | grep -i node-type
[root@k8s01 yaml]#
9. 驱逐节点上的Pod
[root@k8s01 yaml]# kubectl cordon k8s03 --将新创建的pod不能调度到k8s03节点，以前的Pod不受影响。
node/k8s03 cordoned
[root@k8s01 yaml]# kubectl get nodes
NAME STATUS ROLES AGE VERSION
k8s01 Ready master 71d v1.16.0
k8s02 Ready
k8s03 Ready,
SchedulingDisabled
[root@k8s01 yaml]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
pod-02-5c54dc6489-442kx 1/1 Running 0 12m 10.244.2.48
k8s03
pod-02-5c54dc6489-92l8m 1/1 Running 0 12m 10.244.2.49
k8s03
pod-02-5c54dc6489-k4bc7 1/1 Running 0 12m 10.244.0.58 k8s01
[root@k8s01 yaml]# kubectl uncordon k8s03 --释放调度规则
node/k8s03 uncordoned
[root@k8s01 yaml]# kubectl get nodes
NAME STATUS ROLES AGE VERSION
k8s01 Ready master 71d v1.16.0
k8s02 Ready
k8s03 Ready
[root@k8s01 yaml]# kubectl drain k8s03 --ignore-daemonsets --去除k8s03节点上的Pod
node/k8s03 already cordoned
WARNING: ignoring DaemonSet-managed Pods: kube-system/kube-flannel-ds-amd64-cg795, kube-system/kube-proxy-h5dkf
evicting pod "pod-02-5c54dc6489-92l8m"
evicting pod "pod-02-5c54dc6489-442kx"
pod/pod-02-5c54dc6489-92l8m evicted
pod/pod-02-5c54dc6489-442kx evicted
node/k8s03 evicted
[root@k8s01 yaml]# kubectl get nodes
NAME STATUS ROLES AGE VERSION
k8s01 Ready master 71d v1.16.0
k8s02 Ready
k8s03 Ready,
SchedulingDisabled
[root@k8s01 yaml]# kubectl get pods -o wide --以前k8s03节点的Pod全部转移到其它节点上
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
pod-02-5c54dc6489-k4bc7 1/1 Running 0 14m 10.244.0.58
k8s01
pod-02-5c54dc6489-mxk46 1/1 Running 0 25s 10.244.1.46
k8s02
pod-02-5c54dc6489-vmb8l 1/1 Running 0 25s 10.244.1.45
k8s02
[root@k8s01 yaml]# kubectl uncordon k8s03 --恢复调度
node/k8s03 uncordoned
[root@k8s01 yaml]# kubectl get nodes
NAME STATUS ROLES AGE VERSION
k8s01 Ready master 71d v1.16.0
k8s02 Ready
k8s03 Ready
[root@k8s01 yaml]#
当前标题：k8s调度器
新闻来源：http://scjbc.cn/article/pdicig.html
