k8s玩转Pod调度基础
1、hostprot
hostPort是指的主机端口，如果pod调度的节点端口被占用，则新的Pod不会调度到该节点
[root@master231 18-scheduler]# cat 02-scheduler-hostPort.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: deploy-hostport
spec:
replicas: 3
selector:
matchLabels:
apps: v1
template:
metadata:
labels:
apps: v1
spec:
containers:
- name: c1
image: harbor250.oldboyedu.com/oldboyedu-xiuxian/apps:v3
ports:
- containerPort: 80
# 将pod的80端口映射到调度worker节点的90端口。
# 这意味着其他pod如果也想要使用90端口是不可以的，因为端口被占用就无法完成调度！
hostPort: 90
kubectl apply -f 02-scheduler-hostPort.yaml
deployment.apps/deploy-hostport created
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-hostport-565858676b-6ncs4 1/1 Running 0 8s 10.100.1.119 worker232 <none> <none>
deploy-hostport-565858676b-974lr 0/1 Pending 0 8s <none> <none> <none> <none>
deploy-hostport-565858676b-vqhnv 1/1 Running 0 8s 10.100.2.86 worker233 <none> <none>
2、玩转Pod调度基础之hostNetwork
apiVersion: apps/v1
kind: Deployment
metadata:
name: deploy-hostnetwork
spec:
replicas: 3
selector:
matchLabels:
apps: v1
template:
metadata:
labels:
apps: v1
spec:
# 使用宿主机网络通常情况下会搭配dnsPolicy和可选的'containerPort'字段。
hostNetwork: true
dnsPolicy: ClusterFirstWithHostNet
containers:
- name: c1
image: harbor250.oldboyedu.com/oldboyedu-xiuxian/apps:v3
ports:
- containerPort: 80
kubectl apply -f 03-scheduler-hostNetwork.yaml
[root@master231 18-scheduler]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-hostnetwork-68f6858c4b-dkblh 0/1 Pending 0 3s <none> <none> <none> <none>
deploy-hostnetwork-68f6858c4b-mj8n5 1/1 Running 0 3s 10.0.0.232 worker232 <none> <none>
deploy-hostnetwork-68f6858c4b-w6mkw 1/1 Running 0 3s 10.0.0.233 worker233 <none> <none>
3、玩转Pod调度基础之resources
所谓的resources就是配置容器的资源限制。
可以通过requests配置容器的期望资源，可以通过limits配置容器的资源使用上限。
apiVersion: apps/v1
kind: Deployment
metadata:
name: deploy-resources
spec:
replicas: 10
selector:
matchLabels:
app: stress
template:
metadata:
labels:
app: stress
spec:
containers:
# - image: jasonyin2020/oldboyedu-linux-tools:v0.1
- image: harbor250.oldboyedu.com/oldboyedu-test/stress:v0.1
name: oldboyedu-linux-tools
args:
- tail
- -f
- /etc/hosts
# 配置容器的资源限制
resources:
# 1.若节点不满足期望资源，则无法调度到该节点；
# 2.调度到该节点后也不会立刻吃掉期望的资源。
# 3.若没有定义requests字段，则默认等效于limits的配置;
requests:
cpu: 0.5
memory: 300Mi
# memory: 10G
# 1.定义容器资源使用的上限;
# 2.如果没有定义limits资源，则默认使用调度到该节点的所有资源。
limits:
cpu: 1.5
memory: 500Mi
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl apply -f 04-scheduler-resources.yaml
deployment.apps/deploy-resources created
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-resources-7496f54d5b-4jvpz 1/1 Running 0 3m39s 10.100.2.95 worker233 <none> <none>
deploy-resources-7496f54d5b-644wl 1/1 Running 0 3m39s 10.100.1.131 worker232 <none> <none>
deploy-resources-7496f54d5b-67whf 0/1 Pending 0 3m39s <none> <none> <none> <none>
deploy-resources-7496f54d5b-6mfmk 0/1 Pending 0 3m39s <none> <none> <none> <none>
deploy-resources-7496f54d5b-6p86l 0/1 Pending 0 3m39s <none> <none> <none> <none>
deploy-resources-7496f54d5b-dhhcn 0/1 Pending 0 3m39s <none> <none> <none> <none>
deploy-resources-7496f54d5b-h6xxv 1/1 Running 0 3m39s 10.100.2.94 worker233 <none> <none>
deploy-resources-7496f54d5b-qc4vh 1/1 Running 0 3m39s 10.100.2.96 worker233 <none> <none>
deploy-resources-7496f54d5b-rx4nv 1/1 Running 0 3m39s 10.100.1.132 worker232 <none> <none>
deploy-resources-7496f54d5b-skfrd 1/1 Running 0 3m39s 10.100.1.133 worker232 <none> <none>
[root@master231 18-scheduler]#
4、玩转Pod调度基础之Taints
Taints表示污点，作用在worker工作节点上。
污点类型大概分为三类:
- NoSchedule:
不在接受新的Pod调度，已经调度到该节点的Pod不会被驱逐。 - PreferNoSchedule:
优先将Pod调度到其他节点，当其他节点不可调度时，再往该节点调度。 - NoExecute:
不在接受新的Pod调度，且已经调度到该节点的Pod会被立刻驱逐。
污点的格式:
key[=value]:effect
污点测试代码
18-scheduler]# kubectl describe nodes | grep Taints
Taints: node-role.kubernetes.io/master:NoSchedule
Taints: <none>
Taints: <none>
[root@master231 18-scheduler]#
温馨提示:
<none>表示该节点没有污点。
2.2 给指定节点打污点
[root@master231 18-scheduler]# kubectl taint node --all school=oldboyedu:PreferNoSchedule
node/master231 tainted
node/worker232 tainted
node/worker233 tainted
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
school=oldboyedu:PreferNoSchedule
Unschedulable: false
--
Taints: school=oldboyedu:PreferNoSchedule
Unschedulable: false
Lease:
--
Taints: school=oldboyedu:PreferNoSchedule
Unschedulable: false
Lease:
[root@master231 18-scheduler]#
2.3 修改污点【只能修改value字段，修改effect则表示创建了一个新的污点类型】
[root@master231 18-scheduler]# kubectl taint node worker233 school=laonanhai:PreferNoSchedule --overwrite
node/worker233 modified
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
school=oldboyedu:PreferNoSchedule
Unschedulable: false
--
Taints: school=oldboyedu:PreferNoSchedule
Unschedulable: false
Lease:
--
Taints: school=laonanhai:PreferNoSchedule
Unschedulable: false
Lease:
[root@master231 18-scheduler]#
2.3 删除污点
[root@master231 18-scheduler]# kubectl taint node --all school-
node/master231 untainted
node/worker232 untainted
node/worker233 untainted
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
[root@master231 18-scheduler]#
3.测试污点类型案例
3.1 添加污点
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl taint node worker233 school:NoSchedule
node/worker233 tainted
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: school:NoSchedule
Unschedulable: false
Lease:
[root@master231 18-scheduler]#
3.2 测试案例
[root@master231 18-scheduler]# cat 05-scheduler-taints.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: deploy-taints
spec:
replicas: 5
selector:
matchLabels:
app: stress
template:
metadata:
labels:
app: stress
spec:
containers:
- image: harbor250.oldboyedu.com/oldboyedu-test/stress:v0.1
name: oldboyedu-linux-tools
args:
- tail
- -f
- /etc/hosts
resources:
requests:
cpu: 0.5
memory: 300Mi
limits:
cpu: 1.5
memory: 500Mi
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl apply -f 05-scheduler-taints.yaml
deployment.apps/deploy-taints created
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-taints-7496f54d5b-9qfch 0/1 Pending 0 4s <none> <none> <none> <none>
deploy-taints-7496f54d5b-llmqv 1/1 Running 0 4s 10.100.1.138 worker232 <none> <none>
deploy-taints-7496f54d5b-n4vx4 0/1 Pending 0 4s <none> <none> <none> <none>
deploy-taints-7496f54d5b-nvm7h 1/1 Running 0 4s 10.100.1.139 worker232 <none> <none>
deploy-taints-7496f54d5b-t9bxt 1/1 Running 0 4s 10.100.1.137 worker232 <none> <none>
[root@master231 18-scheduler]#
3.3 修改污点类型
[root@master231 18-scheduler]# kubectl taint node worker233 school:PreferNoSchedule
node/worker233 tainted
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl taint node worker233 school:NoSchedule-
node/worker233 untainted
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: school:PreferNoSchedule
Unschedulable: false
Lease:
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-taints-7496f54d5b-9qfch 1/1 Running 0 2m32s 10.100.2.99 worker233 <none> <none>
deploy-taints-7496f54d5b-llmqv 1/1 Running 0 2m32s 10.100.1.138 worker232 <none> <none>
deploy-taints-7496f54d5b-n4vx4 1/1 Running 0 2m32s 10.100.2.100 worker233 <none> <none>
deploy-taints-7496f54d5b-nvm7h 1/1 Running 0 2m32s 10.100.1.139 worker232 <none> <none>
deploy-taints-7496f54d5b-t9bxt 1/1 Running 0 2m32s 10.100.1.137 worker232 <none> <none>
[root@master231 18-scheduler]#
3.4 再次修改污点类型
[root@master231 18-scheduler]# kubectl taint node worker233 school=oldboyedu:NoExecute
node/worker233 tainted
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-taints-7496f54d5b-4rpbg 0/1 Pending 0 1s <none> <none> <none> <none>
deploy-taints-7496f54d5b-9qfch 1/1 Terminating 0 4m35s 10.100.2.99 worker233 <none> <none>
deploy-taints-7496f54d5b-llmqv 1/1 Running 0 4m35s 10.100.1.138 worker232 <none> <none>
deploy-taints-7496f54d5b-n4vx4 1/1 Terminating 0 4m35s 10.100.2.100 worker233 <none> <none>
deploy-taints-7496f54d5b-nvm7h 1/1 Running 0 4m35s 10.100.1.139 worker232 <none> <none>
deploy-taints-7496f54d5b-t2pcv 0/1 Pending 0 1s <none> <none> <none> <none>
deploy-taints-7496f54d5b-t9bxt 1/1 Running 0 4m35s 10.100.1.137 worker232 <none> <none>
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-taints-7496f54d5b-4rpbg 0/1 Pending 0 33s <none> <none> <none> <none>
deploy-taints-7496f54d5b-llmqv 1/1 Running 0 5m7s 10.100.1.138 worker232 <none> <none>
deploy-taints-7496f54d5b-nvm7h 1/1 Running 0 5m7s 10.100.1.139 worker232 <none> <none>
deploy-taints-7496f54d5b-t2pcv 0/1 Pending 0 33s <none> <none> <none> <none>
deploy-taints-7496f54d5b-t9bxt 1/1 Running 0 5m7s 10.100.1.137 worker232 <none> <none>
[root@master231 18-scheduler]#
3.5.删除测试
[root@master231 18-scheduler]# kubectl delete -f 05-scheduler-taints.yaml
deployment.apps "deploy-taints" deleted
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: school=oldboyedu:NoExecute
school:PreferNoSchedule
Unschedulable: false
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl taint node worker233 school-
node/worker233 untainted
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
[root@master231 18-scheduler]#
5、玩转Pod调度基础之tolerations
tolerations是污点容忍，用该技术可以让Pod调度到一个具有污点的节点。
指的注意的是，一个Pod如果想要调度到某个worker节点，则必须容忍该worker的所有污点。
kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
kubectl taint node --all school=oldboyedu:NoSchedule
kubectl taint node worker233 class=linux98:NoExecute
kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
school=oldboyedu:NoSchedule
Unschedulable: false
--
Taints: school=oldboyedu:NoSchedule
Unschedulable: false
Lease:
--
Taints: class=linux98:NoExecute
school=oldboyedu:NoSchedule
Unschedulable: false
cat 06-scheduler-tolerations.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: deploy-tolerations
spec:
replicas: 10
selector:
matchLabels:
app: stress
template:
metadata:
labels:
app: stress
spec:
# 配置污点容忍
tolerations:
# 指定污点的key，如果不定义，则默认匹配所有的key。
- key: school
# 指定污点的value，如果不定义，则默认匹配所有的value。
value: oldboyedu
# 指定污点的effect类型，如果不定义，则默认匹配所有的effect类型。
effect: NoSchedule
- key: class
value: linux98
effect: NoExecute
- key: node-role.kubernetes.io/master
effect: NoSchedule
# # 注意，operator表示key和value的关系，有效值为: Exists and Equal，默认值为: Equal。
# # 如果将operator的值设置为: Exists，且不定义key，value，effect时，表示无视污点。
#- operator: Exists
containers:
- image: harbor250.oldboyedu.com/oldboyedu-test/stress:v0.1
name: oldboyedu-linux-tools
args:
- tail
- -f
- /etc/hosts
resources:
requests:
cpu: 0.5
memory: 300Mi
limits:
cpu: 1.5
memory: 500Mi
[root@master231 18-scheduler]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-tolerations-8667479cb9-85zfb 0/1 Pending 0 4s <none> <none> <none> <none>
deploy-tolerations-8667479cb9-bn7r6 1/1 Running 0 4s 10.100.2.110 worker233 <none> <none>
deploy-tolerations-8667479cb9-d24mp 1/1 Running 0 4s 10.100.1.149 worker232 <none> <none>
deploy-tolerations-8667479cb9-d4gqx 1/1 Running 0 4s 10.100.2.111 worker233 <none> <none>
deploy-tolerations-8667479cb9-dk64w 1/1 Running 0 4s 10.100.2.108 worker233 <none> <none>
deploy-tolerations-8667479cb9-gfqb9 1/1 Running 0 4s 10.100.1.148 worker232 <none> <none>
deploy-tolerations-8667479cb9-n985j 1/1 Running 0 4s 10.100.0.26 master231 <none> <none>
deploy-tolerations-8667479cb9-s88np 1/1 Running 0 4s 10.100.1.147 worker232 <none> <none>
deploy-tolerations-8667479cb9-swnbk 1/1 Running 0 4s 10.100.0.25 master231 <none> <none>
deploy-tolerations-8667479cb9-t52qj 1/1 Running 0 4s 10.100.2.109 worker233 <none> <none>
kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
school=oldboyedu:NoSchedule
Unschedulable: false
--
Taints: school=oldboyedu:NoSchedule
Unschedulable: false
Lease:
--
Taints: class=linux98:NoExecute
school=oldboyedu:NoSchedule
删除名为school 的污点
kubectl taint node --all school-
node/master231 untainted
node/worker232 untainted
node/worker233 untainted
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl taint node worker233 class-
node/worker233 untainted
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
6、玩转Pod调度基础之cordon
cordon标记节点不可调度，一般用于集群维护。
cordon的底层实现逻辑其实就给节点打污点。
kubectl cordon worker233
node/worker233 cordoned
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl get nodes
NAME STATUS ROLES AGE VERSION
master231 Ready control-plane,master 13d v1.23.17
worker232 Ready <none> 13d v1.23.17
worker233 Ready,SchedulingDisabled <none> 13d v1.23.17
[root@master231 18-scheduler]#
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: node.kubernetes.io/unschedulable:NoSchedule
Unschedulable: true
Lease:
kubectl apply -f 05-scheduler-taints.yaml
root@master231 18-scheduler]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-taints-7496f54d5b-5h2xq 1/1 Running 0 6s 10.100.1.153 worker232 <none> <none>
deploy-taints-7496f54d5b-794cp 1/1 Running 0 6s 10.100.1.154 worker232 <none> <none>
deploy-taints-7496f54d5b-jsk5b 1/1 Running 0 6s 10.100.1.152 worker232 <none> <none>
deploy-taints-7496f54d5b-n7q7m 0/1 Pending 0 6s <none> <none> <none> <none>
deploy-taints-7496f54d5b-png9p 0/1 Pending 0 6s <none> <none> <none> <none>
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl delete -f 05-scheduler-taints.yaml
deployment.apps "deploy-taints" deleted
[root@master231 18-scheduler]#
7、玩转Pod调度基础之uncordon
uncordon的操作和cordon操作相反，表示取消节点不可调度功能。
kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: node.kubernetes.io/unschedulable:NoSchedule
Unschedulable: true
Lease:
[root@master231 18-scheduler]#
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl uncordon worker233
node/worker233 uncordoned
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl get nodes
NAME STATUS ROLES AGE VERSION
master231 Ready control-plane,master 13d v1.23.17
worker232 Ready <none> 13d v1.23.17
worker233 Ready <none> 13d v1.23.17
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl apply -f 05-scheduler-taints.yaml
deployment.apps/deploy-taints created
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-taints-7496f54d5b-bfnrh 1/1 Running 0 3s 10.100.2.116 worker233 <none> <none>
deploy-taints-7496f54d5b-csds7 1/1 Running 0 3s 10.100.2.117 worker233 <none> <none>
deploy-taints-7496f54d5b-hqrnm 1/1 Running 0 3s 10.100.2.118 worker233 <none> <none>
deploy-taints-7496f54d5b-k8n66 1/1 Running 0 3s 10.100.1.156 worker232 <none> <none>
deploy-taints-7496f54d5b-pdnkr 1/1 Running 0 3s 10.100.1.155 worker232 <none> <none>
[root@master231 18-scheduler]#
8、玩转Pod调度基础之drain
所谓drain其实就是将所在节点的pod进行驱逐的操作，说白了，就是将当前节点的Pod驱逐到其他节点运行。
在驱逐Pod时，需要忽略ds控制器创建的pod。
驱逐的主要应用场景是集群的缩容。
drain底层调用的cordon。
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-taints-7496f54d5b-bfnrh 1/1 Running 0 4m19s 10.100.2.116 worker233 <none> <none>
deploy-taints-7496f54d5b-csds7 1/1 Running 0 4m19s 10.100.2.117 worker233 <none> <none>
deploy-taints-7496f54d5b-hqrnm 1/1 Running 0 4m19s 10.100.2.118 worker233 <none> <none>
deploy-taints-7496f54d5b-k8n66 1/1 Running 0 4m19s 10.100.1.156 worker232 <none> <none>
deploy-taints-7496f54d5b-pdnkr 1/1 Running 0 4m19s 10.100.1.155 worker232 <none> <none>
[root@master231 18-scheduler]#
2.2 驱逐worker233节点的Pod
[root@master231 18-scheduler]# kubectl drain worker233 --ignore-daemonsets --delete-emptydir-data
node/worker233 already cordoned
WARNING: ignoring DaemonSet-managed Pods: kube-flannel/kube-flannel-ds-6k42r, kube-system/kube-proxy-g5sfd, metallb-system/speaker-8q9ft
evicting pod kube-system/metrics-server-57c6f647bb-54f7w
evicting pod default/deploy-taints-7496f54d5b-csds7
evicting pod default/deploy-taints-7496f54d5b-bfnrh
evicting pod default/deploy-taints-7496f54d5b-hqrnm
pod/metrics-server-57c6f647bb-54f7w evicted
pod/deploy-taints-7496f54d5b-bfnrh evicted
pod/deploy-taints-7496f54d5b-csds7 evicted
pod/deploy-taints-7496f54d5b-hqrnm evicted
node/worker233 drained
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl get pods -o wide # 目前来说，其他节点无法完成调度时则处于Pending状态。
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-taints-7496f54d5b-9mkj5 0/1 Pending 0 58s <none> <none> <none> <none>
deploy-taints-7496f54d5b-k8n66 1/1 Running 0 6m16s 10.100.1.156 worker232 <none> <none>
deploy-taints-7496f54d5b-p2tfw 0/1 Pending 0 58s <none> <none> <none> <none>
deploy-taints-7496f54d5b-pdnkr 1/1 Running 0 6m16s 10.100.1.155 worker232 <none> <none>
deploy-taints-7496f54d5b-vrmfg 1/1 Running 0 58s 10.100.1.157 worker232 <none> <none>
[root@master231 18-scheduler]#
2.3 验证drain底层调用cordon
[root@master231 18-scheduler]# kubectl get nodes
NAME STATUS ROLES AGE VERSION
master231 Ready control-plane,master 13d v1.23.17
worker232 Ready <none> 13d v1.23.17
worker233 Ready,SchedulingDisabled <none> 13d v1.23.17
[root@master231 18-scheduler]#
[root@master231 18-scheduler]# kubectl describe nodes | grep Taints -A 2
Taints: node-role.kubernetes.io/master:NoSchedule
Unschedulable: false
Lease:
--
Taints: <none>
Unschedulable: false
Lease:
--
Taints: node.kubernetes.io/unschedulable:NoSchedule
Unschedulable: true
Lease:
[root@master231 18-scheduler]#
浙公网安备 33010602011771号