概述

在使用过程中，如果希望指定的pod放在自己希望的node节点上可以进行调度，选择对应的node节点。
可以看出可以使用节点的ip地址或者是在node节点添加标签，让pod调度器使用selector进行选择，这里我将使用两种方法，将主服务的beego端放在node1上，将判题器的两个judger放在node2上，judger1使用ip选择，judger2使用lable选择。

调度

nodename调度

apiVersion: apps/v1
kind: Deployment
metadata:name: unioj-judger-deploymentnamespace: default
spec:selector:matchLabels:app: unioj-judgerreplicas: 1template:metadata:labels:app: unioj-judgerspec:nodeName: k8s-node2restartPolicy: Alwayscontainers:- name: unioj-judgerimage: registry.cn-beijing.aliyuncs.com/uyistcoj/unioj-judgerimagePullPolicy: Alwaysports:- containerPort: 7999resources:requests:memory: 100Micpu: 400mlivenessProbe:httpGet:path: /healthyport: 7999initialDelaySeconds: 3periodSeconds: 3successThreshold: 1        # 健康阀值

问题：直接使用192.168.10.66会导致pending一直无法调度，故而使用node的名字，估摸着是集群不认识ip

lable调度

打标签

打标签

kubectl label nodes <node-name> <label-key>=<label-value>

查看标签

[root@k8s-master1 pods]# kubectl get node --show-labels
NAME          STATUS   ROLES                  AGE   VERSION   LABELS
k8s-master1   Ready    control-plane,master   30d   v1.21.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-master1,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=,node.kubernetes.io/exclude-from-external-load-balancers=
k8s-node1     Ready    <none>                 30d   v1.21.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node1,kubernetes.io/os=linux
k8s-node2     Ready    <none>                 30d   v1.21.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node2,kubernetes.io/os=linux

给node2打标签unioj-role: judgers

[root@k8s-master1 pods]# kubectl label nodes k8s-node2 unioj-role=judgers
node/k8s-node2 labeled
[root@k8s-master1 pods]# kubectl get node --show-labels
NAME          STATUS   ROLES                  AGE   VERSION   LABELS
k8s-master1   Ready    control-plane,master   30d   v1.21.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-master1,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=,node.kubernetes.io/exclude-from-external-load-balancers=
k8s-node1     Ready    <none>                 30d   v1.21.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node1,kubernetes.io/os=linux
k8s-node2     Ready    <none>                 30d   v1.21.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node2,kubernetes.io/os=linux,unioj-role=judgers

修改yaml进行选择

apiVersion: apps/v1
kind: Deployment
metadata:name: unioj-judger-deployment2namespace: default
spec:selector:matchLabels:app: unioj-judger2replicas: 1template:metadata:labels:app: unioj-judger2spec:nodeSelector:unioj-role: judgerscontainers:- name: unioj-judgerimage: registry.cn-beijing.aliyuncs.com/uyistcoj/unioj-judgerimagePullPolicy: Alwaysports:- containerPort: 7999resources:requests:memory: 100Micpu: 400mimagePullPolicy: Alwaysports:- containerPort: 7999resources:requests:memory: 100Micpu: 400mlivenessProbe:httpGet:path: /healthyport: 7999initialDelaySeconds: 3periodSeconds: 3successThreshold: 1        # 健康阀值

apply之前的分布

[root@k8s-master1 pods]# kubectl get pods -o wide
NAME                                        READY   STATUS    RESTARTS   AGE     IP            NODE        NOMINATED NODE   READINESS GATES
unioj-beego-deployment-6874449649-lhjm7     1/1     Running   1          11h     10.100.2.64   k8s-node2   <none>           <none>
unioj-judger-deployment-5c596bb8d5-gck6f    1/1     Running   0          7m34s   10.100.2.68   k8s-node2   <none>           <none>
unioj-judger-deployment2-78ff65b5fd-99r5j   1/1     Running   0          46m     10.100.1.51   k8s-node1   <none>           <none>

apply

[root@k8s-master1 pods]# kubectl apply -f unioj-judger2.yaml
deployment.apps/unioj-judger-deployment2 configured[root@k8s-master1 pods]# kubectl get pods -o wide
NAME                                        READY   STATUS    RESTARTS   AGE     IP            NODE        NOMINATED NODE   READINESS GATES
unioj-beego-deployment-6874449649-lhjm7     1/1     Running   1          11h     10.100.2.64   k8s-node2   <none>           <none>
unioj-judger-deployment-5c596bb8d5-gck6f    1/1     Running   0          8m48s   10.100.2.68   k8s-node2   <none>           <none>
unioj-judger-deployment2-5fcf476c57-xf8ck   1/1     Running   0          27s     10.100.2.69   k8s-node2   <none>           <none>

同样的方法把beego调度到node1

[root@k8s-master1 pods]# kubectl label node k8s-node1 unioj-role=beego
node/k8s-node1 labeled[root@k8s-master1 pods]# kubectl get node --show-labels
NAME          STATUS   ROLES                  AGE   VERSION   LABELS
k8s-master1   Ready    control-plane,master   30d   v1.21.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-master1,kubernetes.io/os=linux,node-role.kubernetes.io/control-plane=,node-role.kubernetes.io/master=,node.kubernetes.io/exclude-from-external-load-balancers=
k8s-node1     Ready    <none>                 30d   v1.21.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node1,kubernetes.io/os=linux,unioj-role=beego
k8s-node2     Ready    <none>                 30d   v1.21.2   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=k8s-node2,kubernetes.io/os=linux,unioj-role=judgers

yaml

apiVersion: apps/v1
kind: Deployment
metadata:name: unioj-beego-deploymentnamespace: defaultlabels:app: unioj-beego
spec:replicas: 1selector:matchLabels:app: unioj-beegotemplate:metadata:labels:app: unioj-beegospec:nodeSelector:unioj-role: beegocontainers:- name: unoj-beegoimage: registry.cn-beijing.aliyuncs.com/uyistcoj/unioj-beego:latestports:- containerPort: 8000- containerPort: 8666

apply后

[root@k8s-master1 pods]# kubectl get pod -o wide
NAME                                        READY   STATUS    RESTARTS   AGE     IP            NODE        NOMINATED NODE   READINESS GATES
unioj-beego-deployment-dc8986bcb-7mrqv      1/1     Running   0          4m13s   10.100.1.52   k8s-node1   <none>           <none>
unioj-judger-deployment-5c596bb8d5-gck6f    1/1     Running   0          15m     10.100.2.68   k8s-node2   <none>           <none>
unioj-judger-deployment2-5fcf476c57-xf8ck   1/1     Running   0          7m      10.100.2.69   k8s-node2   <none>           <none>

结果

所有的judger到node2上，主控的beego到了node1

常见问题

k8s指定node调度相关推荐

k8s 指定节点调度标签
nodeSelector 强制匹配指定pod 调度到标签 ssd [root@k8s-master cka3]# cat ssd.yaml apiVersion: v1 kind: Pod meta ...
关于K8s中Pod调度[选择器,指定节点,主机亲和性]方式和节点[coedon,drain,taint]标记的Demo
写在前面嗯,整理K8s中pod调度相关笔记,这里分享给小伙伴博文内容涉及: kube-scheduler组件的简述 Pod的调度(选择器.指定节点.主机亲和性)方式节点的coedon与drain ...
k8s 部署pod到指定node
k8s指定节点部署在一些业务场景中,会需要将一些pod部署到指定node,按照默认的调度规则,pod会优先分配到负载较小的node中,难免会出现多个pod资源竞争的情况. k8s有两种常用的方法可以 ...
k8s ready 不调度_【零基础学云计算】k8s部署---master节点组件部署（三）
kube-APIserver组件介绍 kube-APIserver提供了k8s各类资源对象(pod,RC,Service等)的增删改查及watch等HTTP Rest接口,是整个系统的数据总线和数据中 ...
K8S之Scheduler调度器
K8S之Scheduler调度器 kubernetes Scheduler 简介 kubernetes Scheduler 运行在 master 节点,它的核心功能是监听 apiserver 来获取 ...
k8s(八）—调度因素（nodeName、nodeSelector、亲和与反亲和、Taints 污点）、影响pod调度的指令
1 调度简介 [root@server2 ~]# kubectl get pod -n kube-system 2 影响kubernetes调度的因素 2.1 nodeName(针对节点) [root ...
OpenShift 4 - 通过DaemonSet在指定Node上运行守护程序
<OpenShift 4.x HOL教程汇总> 文章目录通过DaemonSet在Worker Node上运行守护程序通过DaemonSet在指定Node上运行守护程序通过Daemon ...
nvm安装流程、使用nvm安装指定node版本
注意:要用nvm管理node版本的前提是要把之前已经下载的node给卸载掉,然后用nvm陆续添加自己所需的node版本,否则会出现nvm中node版本切换过来了,但是再使用的时候还是原来的. 1.下载 ...
Mac 指定node版本安装
Mac 指定node版本安装之前一直没有写文章的习惯,经过几次处理技术问题的事情后,发现还是要花点时间记录一下日常遇到的一些问题及解决办法,有助于加深印象和后续再次遇到时,能快速的回忆起来 1.nv ...

k8s指定node调度

概述

调度

nodename调度

lable调度

打标签

结果

常见问题

k8s指定node调度相关推荐

最新文章

热门文章