pod 的親和性，反親和性 實驗
阿新 • • 發佈：2021-11-09
pod 的親和性，反親和性 實驗
目錄1 ,環境準備
node01、 node02 都有標籤 test=a，有個pod1 執行在node01上， 標籤為app=myapp01
#設定node01和node02節點，擁有標籤 test=a [root@master demo]# kubectl label nodes node{01,02} test=a --overwrite node/node01 labeled node/node02 labeled #檢視擁有標籤test=a的節點 [root@master demo]# kubectl get nodes -l test=a NAME STATUS ROLES AGE VERSION node01 Ready <none> 7d6h v1.15.1 node02 Ready <none> 7d6h v1.15.1
[root@master demo]# vim test.yaml
apiVersion: v1
kind: Pod
metadata:
name: myapp01
labels:
app: myapp01
spec:
containers:
- name: with-node-affinity
image: soscscs/myapp:v1
#宣告式建立pod [root@master demo]# kubectl apply -f test.yaml pod/myapp01 created #檢視pod myapp01的詳細資訊的標籤 [root@master demo]# kubectl get pods myapp01 -o wide --show-labels NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS myapp01 1/1 Running 0 26s 10.244.1.112 node01 <none> <none> app=myapp01
2 親和性+ In 測試
[root@master demo]# vim a.yaml apiVersion: v1 kind: Pod metadata: name: myapp10 labels: app: myapp03 spec: containers: - name: myapp03 image: soscscs/myapp:v1 affinity: podAffinity: requiredDuringSchedulingIgnoredDuringExecution: - labelSelector: matchExpressions: - key: app operator: In values: - myapp01 topologyKey: test
pod擁有app=myapp01 標籤（設這個pod為x），則把新pod排程到和 pod x 擁有同一個拓撲域test=a 的 節點上。
[root@master demo]# for i in myapp{11..15}; do kubectl apply -f a.yaml; sed -ri "4s#myapp1[0-9]#$i#" a.yaml; done
pod/myapp10 created
pod/myapp11 created
pod/myapp12 created
pod/myapp13 created
pod/myapp14 created
[root@master demo]# kubectl get pods -o wide --show-labels
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
myapp01 1/1 Running 0 88m 10.244.1.112 node01 <none> <none> app=myapp01
myapp10 1/1 Running 0 6s 10.244.2.104 node02 <none> <none> app=myapp03
myapp11 1/1 Running 0 6s 10.244.1.138 node01 <none> <none> app=myapp03
myapp12 1/1 Running 0 5s 10.244.2.105 node02 <none> <none> app=myapp03
myapp13 1/1 Running 0 5s 10.244.1.139 node01 <none> <none> app=myapp03
myapp14 1/1 Running 0 5s 10.244.2.106 node02 <none> <none> app=myapp03
因為拓撲域test=a內，已經有了pod存在於node01節點上，所以
3 親和性+NotIn 測試
[root@master demo]# vim a.yaml
apiVersion: v1
kind: Pod
metadata:
name: myapp10
labels:
app: myapp03
spec:
containers:
- name: myapp03
image: soscscs/myapp:v1
affinity:
podAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchExpressions:
- key: app
operator: NotIn
values:
- myapp01
topologyKey: test
[root@master demo]# kubectl delete pod myapp{10..14};for i in myapp{11..15}; do kubectl apply -f a.yaml; sed -ri "4s#myapp1[0-9]#$i#" a.yaml; done
pod "myapp10" deleted
pod "myapp11" deleted
pod "myapp12" deleted
pod "myapp13" deleted
pod "myapp14" deleted
pod/myapp10 created
pod/myapp11 created
pod/myapp12 created
pod/myapp13 created
pod/myapp14 created
[root@master demo]# kubectl get pods -o wide --show-labels
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
myapp01 1/1 Running 0 91m 10.244.1.112 node01 <none> <none> app=myapp01
myapp10 1/1 Running 0 5s 10.244.1.140 node01 <none> <none> app=myapp03
myapp11 1/1 Running 0 5s 10.244.2.107 node02 <none> <none> app=myapp03
myapp12 1/1 Running 0 4s 10.244.1.141 node01 <none> <none> app=myapp03
myapp13 1/1 Running 0 4s 10.244.2.108 node02 <none> <none> app=myapp03
myapp14 1/1 Running 0 4s 10.244.1.142 node01 <none> <none> app=myapp03
4 非親和性+In 測試
[root@master demo]# vim a.yaml
apiVersion: v1
kind: Pod
metadata:
name: myapp10
labels:
app: myapp03
spec:
containers:
- name: myapp03
image: soscscs/myapp:v1
affinity:
podAntiAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchExpressions:
- key: app
operator: In
values:
- myapp01
topologyKey: test
[root@master demo]# kubectl delete pod myapp{10..14};for i in myapp{11..15}; do kubectl apply -f a.yaml; sed -ri "4s#myapp1[0-9]#$i#" a.yaml; done
pod "myapp10" deleted
pod "myapp11" deleted
pod "myapp12" deleted
pod "myapp13" deleted
pod "myapp14" deleted
pod/myapp10 created
pod/myapp11 created
pod/myapp12 created
pod/myapp13 created
pod/myapp14 created
[root@master demo]# kubectl get pods -o wide --show-labels
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
myapp01 1/1 Running 0 96m 10.244.1.112 node01 <none> <none> app=myapp01
myapp10 0/1 Pending 0 4s <none> <none> <none> <none> app=myapp03
myapp11 0/1 Pending 0 3s <none> <none> <none> <none> app=myapp03
myapp12 0/1 Pending 0 3s <none> <none> <none> <none> app=myapp03
myapp13 0/1 Pending 0 3s <none> <none> <none> <none> app=myapp03
myapp14 0/1 Pending 0 3s <none> <none> <none> <none> app=myapp03
[root@master demo]# kubectl describe pod myapp10
5 非親和性 + NotIn 測試
[root@master demo]# vim a.yaml
apiVersion: v1
kind: Pod
metadata:
name: myapp10
labels:
app: myapp03
spec:
containers:
- name: myapp03
image: soscscs/myapp:v1
affinity:
podAntiAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchExpressions:
- key: app
operator: NotIn
values:
- myapp01
topologyKey: test
[root@master demo]# kubectl delete pod myapp{10..14};for i in myapp{11..15}; do kubectl apply -f a.yaml; sed -ri "4s#myapp1[0-9]#$i#" a.yaml; done
pod "myapp10" deleted
pod "myapp11" deleted
pod "myapp12" deleted
pod "myapp13" deleted
pod "myapp14" deleted
pod/myapp10 created
pod/myapp11 created
pod/myapp12 created
pod/myapp13 created
pod/myapp14 created
[root@master demo]# kubectl get pods -o wide --show-labels
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES LABELS
myapp01 1/1 Running 0 102m 10.244.1.112 node01 <none> <none> app=myapp01
myapp10 1/1 Running 0 8s 10.244.2.109 node02 <none> <none> app=myapp03
myapp11 0/1 Pending 0 8s <none> <none> <none> <none> app=myapp03
myapp12 0/1 Pending 0 7s <none> <none> <none> <none> app=myapp03
myapp13 0/1 Pending 0 7s <none> <none> <none> <none> app=myapp03
myapp14 0/1 Pending 0 7s <none> <none> <none> <none> app=myapp03
[root@master demo]# kubectl describe pod myapp11