kubernetes核心概念 Controller
pod控制器controller
Controller作用及分类
controller用于控制pod
参考: kubernetes.io/zh/docs/con…
控制器主要分为:
- Deployments 部署无状态应用，控制pod升级，回退
- ReplicaSet 副本集，控制pod扩容，裁减
- ReplicationController(相当于ReplicaSet的老版本，现在建议使用Deployments加ReplicaSet替代RC)
- StatefulSets 部署有状态应用，结合Service、存储等实现对有状态应用部署
- DaemonSet 守护进程集，运行在所有集群节点(包括master)， 比如使用filebeat，node_exporter
- Jobs 一次性
- Cronjob 周期性
Deployment
Replicaset控制器的功能
- 支持新的基于集合的selector(以前的rc里没有这种功能)
- 通过改变Pod副本数量实现Pod的扩容和缩容
Deployment控制器的功能
- Deployment集成了上线部署、滚动升级、创建副本、回滚等功能
- Deployment里包含并使用了ReplicaSet
Deployment用于部署无状态应用
无状态应用的特点:
- 所有pod无差别
- 所有pod中容器运行同一个image
- 所有pod可以运行在集群中任意node上
- 所有pod无启动顺序先后之分
- 随意pod数量扩容或缩容
- 例如简单运行一个静态web程序
创建deployment类型应用
deployment-nginx.yml
apiVersion: apps/v1
kind: Deployment
metadata:
name: deploy-nginx # deployment名
spec:
replicas: 1 # 副本集,deployment里使用了replicaset
selector:
matchLabels:
app: nginx # 匹配的pod标签,表示deployment和rs控制器控制带有此标签的pod
template: # 代表pod的配置模板
metadata:
labels:
app: nginx # pod的标签
spec:
containers: # 以下为pod里的容器定义
- name: nginx
image: nginx:1.15-alpine
imagePullPolicy: IfNotPresent
ports:
- containerPort: 80
# 应用YAML文件创建deployment
kubectl apply -f deployment-nginx.yml
# 也可以通过命令的方式
# kubectl create deployment deployment名称 --image=镜像名称 --replicas 集群个数
# 查看验证，deployment可简写成deploy
kubectl get deployment
NAME READY UP-TO-DATE AVAILABLE AGE
deploy-nginx 1/1 1 1 32s
kubectl get pods
NAME READY STATUS RESTARTS AGE
deploy-nginx-79f57dcc-cxq9p 1/1 Running 0 92s
# replicasets可简写成rs
kubectl get replicasets
NAME DESIRED CURRENT READY AGE
deploy-nginx-79f57dcc 1 1 1 2m2s
访问deployment
# 查看pod的IP地址
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-nginx-79f57dcc-cxq9p 1/1 Running 0 14h 10.244.30.89 worker02 <none> <none>
# 在任意集群节点上都可以访问此deploy里pod，但集群外部是不能访问的
curl 10.244.30.89
<!DOCTYPE html>
<html>
<head>
<title>Welcome to nginx!</title>
...
删除deployment中的pod
# 删除pod（注意: 是删除deployment中的pod）
kubectl delete pod deploy-nginx-79f57dcc-cxq9p
# 再次查看，发现又重新启动了一个pod(节点由worker02转为worker01了，IP地址也变化了)
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-nginx-79f57dcc-qr9j4 1/1 Running 0 27s 10.244.5.13 worker01 <none> <none>
pod的IP不是固定的，比如把整个集群关闭再启动，pod也会自动启动，但是IP地址也会变化。既然IP地址不是固定的，所以需要一个固定的访问endpoint给用户，那么这种方式就是service
pod版本升级
# 查看帮助
kubectl set image -h
# 升级前验证nginx版本
kubectl describe pods deploy-nginx-79f57dcc-qr9j4 | grep Image:
Image: nginx:1.15-alpine
kubectl exec deploy-nginx-79f57dcc-qr9j4 -- nginx -v
nginx version: nginx/1.15.12
# 升级为1.16版
# 'deployment deploy-nginx'代表名为deploy-nginx的deployment，'nginx=nginx:1.16-alpine'前面的nginx为容器名，--record 表示会记录该选项已经废除
kubectl set image deployment deploy-nginx nginx=nginx:1.16-alpine --record=true
deployment.apps/deploy-nginx image updated
# pod的名称变了
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-nginx-5cd7f4857d-l6stp 1/1 Running 0 7m47s 10.244.30.90 worker02 <none> <none>
# 容器名怎么查看?
kubectl describe pod deploy-nginx-5cd7f4857d-l6stp
kubectl edit deployment deploy-nginx
kubectl get deployment deploy-nginx -o yaml
# 验证
# 如果升级的pod数量较多，则需要一定时间，可通过下面命令查看是否已经成功
kubectl rollout status deployment deploy-nginx
deployment "deploy-nginx" successfully rolled out
# 验证版本，升级为1.16了
kubectl describe pod deploy-nginx-5cd7f4857d-l6stp | grep Image:
Image: nginx:1.16-alpine
kubectl exec deploy-nginx-5cd7f4857d-l6stp -- nginx -v
nginx version: nginx/1.16.1
# 再将nginx1升级为1.17版
kubectl set image deployment deploy-nginx nginx=nginx:1.17-alpine --record=true
deployment.apps/deploy-nginx image updated
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-nginx-77495c487f-88jdl 1/1 Running 0 6s 10.244.30.95 worker02 <none> <none>
kubectl describe pod deploy-nginx-77495c487f-88jdl | grep Image:
Image: nginx:1.17-alpine
pod版本回退
# 查看版本历史信息，最原始是1.15版
kubectl rollout history deployment deploy-nginxkubectl rollout history deployment deploy-nginx
deployment.apps/deploy-nginx
REVISION CHANGE-CAUSE
1 <none>
2 kubectl set image deployment deploy-nginx nginx=nginx:1.16-alpine --record=true
3 kubectl set image deployment deploy-nginx nginx=nginx:1.17-alpine --record=true
# 查看要回退的版本
kubectl rollout history deployment deploy-nginx --revision=1
deployment.apps/deploy-nginx with revision #1
Pod Template:
Labels: app=nginx
pod-template-hash=79f57dcc
Containers:
nginx:
Image: nginx:1.15-alpine
Port: 80/TCP
Host Port: 0/TCP
Environment: <none>
Mounts: <none>
Volumes: <none>
# 执行回退
kubectl rollout undo deployment deploy-nginx --to-revision=1
deployment.apps/deploy-nginx rolled back
# 验证
kubectl rollout history deployment deploy-nginx
deployment.apps/deploy-nginx
REVISION CHANGE-CAUSE
2 kubectl set image deployment deploy-nginx nginx=nginx:1.16-alpine --record=true
3 kubectl set image deployment deploy-nginx nginx=nginx:1.17-alpine --record=true
4 <none>
kubectl get pods
NAME READY STATUS RESTARTS AGE
deploy-nginx-79f57dcc-wpc6m 1/1 Running 0 70s
kubectl describe pod deploy-nginx-79f57dcc-wpc6m | grep Image:
Image: nginx:1.15-alpine
kubectl exec deploy-nginx-79f57dcc-wpc6m -- nginx -v
nginx version: nginx/1.15.12
副本扩容
# 查看帮助
kubectl scale -h
# 扩容为2个副本
kubectl scale deployment deploy-nginx --replicas=2
deployment.apps/deploy-nginx scaled
# 也可以通过修改yaml文件的方式，kubectl edit deploy Deployment名称
# 在两个worker节点上各1个pod
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-nginx-79f57dcc-hvjr5 1/1 Running 0 22s 10.244.30.99 worker02 <none> <none>
deploy-nginx-79f57dcc-wpc6m 1/1 Running 0 7m29s 10.244.5.17 worker01 <none> <none>
# 继续扩容(我们这里只有2个worker，但是可以大于worker节点数)
kubectl scale deployment deploy-nginx --replicas=4
deployment.apps/deploy-nginx scaled
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-nginx-79f57dcc-dmzc8 1/1 Running 0 24s 10.244.5.18 worker01 <none> <none>
deploy-nginx-79f57dcc-hvjr5 1/1 Running 0 3m39s 10.244.30.99 worker02 <none> <none>
deploy-nginx-79f57dcc-mxvg5 1/1 Running 0 24s 10.244.30.100 worker02 <none> <none>
deploy-nginx-79f57dcc-wpc6m 1/1 Running 0 10m 10.244.5.17 worker01 <none> <none>
副本裁减
# 指定副本数为1进行裁减
kubectl scale deployment deploy-nginx --replicas=1
deployment.apps/deploy-nginx scaled
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-nginx-79f57dcc-wpc6m 1/1 Running 0 12m 10.244.5.17 worker01 <none> <none>
多副本滚动更新
# 先扩容副本
kubectl scale deployment deploy-nginx --replicas=16
kubectl get pods
NAME READY STATUS RESTARTS AGE
deploy-nginx-79f57dcc-2xkqt 1/1 Running 0 12s
deploy-nginx-79f57dcc-4h9vq 1/1 Running 0 12s
deploy-nginx-79f57dcc-4tvm7 1/1 Running 0 12s
deploy-nginx-79f57dcc-57hsw 1/1 Running 0 12s
deploy-nginx-79f57dcc-65b8s 1/1 Running 0 12s
deploy-nginx-79f57dcc-bnkfm 1/1 Running 0 12s
deploy-nginx-79f57dcc-f4rq4 1/1 Running 0 12s
deploy-nginx-79f57dcc-fgkl6 1/1 Running 0 12s
deploy-nginx-79f57dcc-fj2nf 1/1 Running 0 12s
deploy-nginx-79f57dcc-g8lm2 1/1 Running 0 12s
deploy-nginx-79f57dcc-lh9h8 1/1 Running 0 12s
deploy-nginx-79f57dcc-n9fwx 1/1 Running 0 12s
deploy-nginx-79f57dcc-nr4ss 1/1 Running 0 12s
deploy-nginx-79f57dcc-rgtq5 1/1 Running 0 12s
deploy-nginx-79f57dcc-vg7bk 1/1 Running 0 12s
deploy-nginx-79f57dcc-wpc6m 1/1 Running 0 13m
# 滚动更新
kubectl set image deployment deploy-nginx nginx=nginx:1.17-alpine --record
deployment.apps/deploy-nginx image updated
# 验证
kubectl rollout status deployment deploy-nginx
...
Waiting for deployment "deploy-nginx" rollout to finish: 12 of 16 updated replicas are available...
Waiting for deployment "deploy-nginx" rollout to finish: 13 of 16 updated replicas are available...
Waiting for deployment "deploy-nginx" rollout to finish: 14 of 16 updated replicas are available...
Waiting for deployment "deploy-nginx" rollout to finish: 15 of 16 updated replicas are available...
deployment "deploy-nginx" successfully rolled out
删除deployment
# 删除deployment，那么里面的pod也会被自动删除
# 直接命令删除
kubectl delete deployment deploy-nginx
# 通过文件的方式删除
kubectl delete -f deployment-nginx.yml
Replicaset
rs-nginx.yml
apiVersion: apps/v1
kind: ReplicaSet
metadata:
name: rs-nginx
namespace: default
spec: # replicaset的spec
replicas: 2 # 副本数
selector: # 标签选择器,对应pod的标签
matchLabels:
app: nginx # 匹配的label
template:
metadata:
name: nginx # pod名
labels: # 对应上面定义的标签选择器selector里面的内容
app: nginx
spec: # pod的spec
containers:
- name: nginx
image: nginx:1.15-alpine
ports:
- name: http
containerPort: 80
# 应用YAML文件
kubectl apply -f rs-nginx.yml
# 验证
kubectl get rs
NAME DESIRED CURRENT READY AGE
rs-nginx 2 2 2 11s
kubectl get pods
NAME READY STATUS RESTARTS AGE
rs-nginx-5mnw5 1/1 Running 0 25s
rs-nginx-chkv5 1/1 Running 0 25s
# 找不到deployment,说明创建rs并没有创建deployment
kubectl get deployment
No resources found in default namespace.
# 删除
kubectl delete -f rs-nginx.yml
pod控制器Controller进阶
DaemonSet
DaemonSet介绍
- DaemonSet能够让所有（或者特定）的节点运行同一个pod
- 当节点加入到K8S集群中，pod会被（DaemonSet）调度到该节点上运行，当节点从K8S集群中被移除，被DaemonSet调度的pod会被移除
- 如果删除DaemonSet，所有跟这个DaemonSet相关的pods都会被删除
- 如果一个DaemonSet的Pod被杀死、停止、或者崩溃，那么DaemonSet将会重新创建一个新的副本在这台计算节点上
- DaemonSet一般应用于日志收集、监控采集、分布式存储守护进程等
DaemonSet应用案例
daemonset-nginx.yml
apiVersion: apps/v1
kind: DaemonSet
metadata:
name: daemonset-nginx
spec:
selector:
matchLabels:
name: nginx-ds
template:
metadata:
labels:
name: nginx-ds
spec:
tolerations: # tolerations代表容忍
- key: node-role.kubernetes.io/master # 能容忍的污点key
effect: NoSchedule # kubectl explain pod.spec.tolerations查看(能容忍的污点effect)
containers:
- name: nginx
image: nginx:1.15-alpine
imagePullPolicy: IfNotPresent
resources: # resources资源限制是为了防止master节点的资源被占太多(根据实际情况配置)
limits:
memory: 100Mi
requests:
memory: 100Mi
# apply应用YAML文件
kubectl apply -f daemonset-nginx.yml
# 验证，daemonset可简写为ds
kubectl get daemonset
NAME DESIRED CURRENT READY UP-TO-DATE AVAILABLE NODE SELECTOR AGE
daemonset-nginx 2 2 2 2 2 <none> 14s
# 每个worker节点都会运行一个pod
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
daemonset-nginx-j4tl5 1/1 Running 0 81s 10.244.30.118 worker02 <none> <none>
daemonset-nginx-vddj2 1/1 Running 0 81s 10.244.5.35 worker01 <none> <none>
# 删除
kubectl delete -f daemonset-nginx.yml
Job
Job介绍
- 对于ReplicaSet而言，它希望pod保持预期数目、持久运行下去，除非用户明确删除，否则这些对象一直存在，它们针对的是耐久性任务，如web服务等
- 对于非耐久性任务，比如压缩文件，任务完成后，pod需要结束运行，不需要pod继续保持在系统中，这个时候就要用到Job
- Job负责批量处理短暂的一次性任务 (short lived one-off tasks)，即仅执行一次的任务，它保证批处理任务的一个或多个Pod成功结束
Job应用案例
计算圆周率2000位
job1.yml
apiVersion: batch/v1
kind: Job
metadata:
name: pi # job名
spec:
template:
metadata:
name: pi # pod名
spec:
containers:
- name: pi # 容器名
image: perl # 此镜像有800多M,可提前导入到所有节点,也可能指定导入到某一节点然后指定调度到此节点
imagePullPolicy: IfNotPresent
command: ["perl", "-Mbignum=bpi", "-wle", "print bpi(2000)"]
restartPolicy: Never # 执行完后不再重启
# 应用YAML文件创建job
kubectl apply -f job1.yml
# 验证，等待COMPLETIONS变为1/1
kubectl get jobs
NAME COMPLETIONS DURATION AGE
pi 1/1 74s 76s
# Completed状态,不再是ready状态
kubectl get pods
NAME READY STATUS RESTARTS AGE
pi-sjzlh 0/1 Completed 0 81s
kubectl logs pi-sjzlh
3.1415926535897932384626433832795028841971693993751058209749445923078164062862089986280348...
# 删除
kubectl delete -f job1.yml
创建固定次数job
job2.yml
apiVersion: batch/v1
kind: Job
metadata:
name: busybox-job
spec:
completions: 10 # 执行job的次数
parallelism: 1 # 执行job的并发数
template:
metadata:
name: busybox-job-pod
spec:
containers:
- name: busybox
image: busybox
imagePullPolicy: IfNotPresent
command: ["echo", "hello"]
restartPolicy: Never
# 应用YAML文件创建job
kubectl apply -f job2.yml
# 验证
kubectl get job
NAME COMPLETIONS DURATION AGE
busybox-job 1/10 4s 5s
# 32秒左右结束
kubectl get job
NAME COMPLETIONS DURATION AGE
busybox-job 10/10 32s 33s
kubectl get pods
NAME READY STATUS RESTARTS AGE
busybox-job-5gb7h 0/1 Completed 0 74s
busybox-job-8brbl 0/1 Completed 0 68s
busybox-job-cqr6t 0/1 Completed 0 62s
busybox-job-ctjx9 0/1 Completed 0 81s
busybox-job-lzfws 0/1 Completed 0 65s
busybox-job-p9rzv 0/1 Completed 0 84s
busybox-job-pqkkd 0/1 Completed 0 77s
busybox-job-rp95t 0/1 Completed 0 55s
busybox-job-s56gr 0/1 Completed 0 71s
busybox-job-tmnx9 0/1 Completed 0 59s
# 删除
kubectl delete -f job2.yml
一次性备份MySQL数据库
通过Job控制器创建应用备份MySQL数据库
MySQL数据库准备
00_mysql.yaml
apiVersion: v1
kind: Service
metadata:
name: mysql-test
namespace: default
spec:
ports:
- port: 3306
name: mysql
clusterIP: None
selector:
app: mysql-dump
---
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: db
namespace: default
spec:
selector:
matchLabels:
app: mysql-dump
serviceName: "mysql-test"
template:
metadata:
labels:
app: mysql-dump
spec:
nodeName: worker01
containers:
- name: mysql
image: mysql:5.7
env:
- name: MYSQL_ROOT_PASSWORD
value: "abc123"
ports:
- containerPort: 3306
volumeMounts:
- mountPath: "/var/lib/mysql"
name: mysql-data
volumes:
- name: mysql-data
hostPath:
path: /opt/mysqldata
创建用于实现任务的资源清单文件
03_job.yaml
apiVersion: batch/v1
kind: Job
metadata:
name: mysql-dump
spec:
template:
metadata:
name: mysql-dump
spec:
nodeName: worker02
containers:
- name: mysql-dump
image: mysql:5.7
command: ["/bin/sh","-c","mysqldump --host=mysql-test -uroot -pabc123 --databases mysql > /root/mysql2023.sql"]
volumeMounts:
- mountPath: "/root"
name: mysql-data
restartPolicy: Never
volumes:
- name: mysql-data
hostPath:
path: /opt/mysqldump
kubectl apply -f 00_mysql.yaml
service/mysql-test created
statefulset.apps/db created
kubectl apply -f 03_job.yaml
job.batch/mysql-dump created
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
db-0 1/1 Running 0 6m47s 10.244.5.36 worker01 <none> <none>
mysql-dump-blxsl 0/1 Completed 0 35s 10.244.30.67 worker02 <none> <none>
# worker02
cat /opt/mysqldump/mysql2023.sql | tail -26
CREATE TABLE IF NOT EXISTS `slow_log` (
`start_time` timestamp(6) NOT NULL DEFAULT CURRENT_TIMESTAMP(6) ON UPDATE CURRENT_TIMESTAMP(6),
`user_host` mediumtext NOT NULL,
`query_time` time(6) NOT NULL,
`lock_time` time(6) NOT NULL,
`rows_sent` int(11) NOT NULL,
`rows_examined` int(11) NOT NULL,
`db` varchar(512) NOT NULL,
`last_insert_id` int(11) NOT NULL,
`insert_id` int(11) NOT NULL,
`server_id` int(10) unsigned NOT NULL,
`sql_text` mediumblob NOT NULL,
`thread_id` bigint(21) unsigned NOT NULL
) ENGINE=CSV DEFAULT CHARSET=utf8 COMMENT='Slow log';
...
# 删除
kubectl delete -f 03_job.yaml
kubectl delete -f 00_mysql.yaml
CronJob
CronJob介绍
- 类似于Linux系统的crontab，在指定的时间周期运行相关的任务
- 时间格式：分时日月周
CronJob应用案例
周期性输出字符
cronjob.yml
apiVersion: batch/v1
kind: CronJob
metadata:
name: cronjob1
spec:
schedule: "* * * * *" # 分时日月周
jobTemplate:
spec:
template:
spec:
containers:
- name: hello
image: busybox
args:
- /bin/sh
- -c
- date; echo hello kubernetes
imagePullPolicy: IfNotPresent
restartPolicy: OnFailure
# 应用YAML文件创建cronjob
kubectl apply -f cronjob.yml
# 查看验证
kubectl get cronjob
NAME SCHEDULE SUSPEND ACTIVE LAST SCHEDULE AGE
cronjob1 * * * * * False 0 <none> 8s
# 看AGE时间,每分钟整点执行一次
kubectl get pods
NAME READY STATUS RESTARTS AGE
cronjob1-28322093-ntxkl 0/1 Completed 0 2m32s
cronjob1-28322094-gr249 0/1 Completed 0 92s
cronjob1-28322095-twqsr 0/1 Completed 0 32s
# 删除
kubectl delete -f cronjob.yml
周期性备份MySQL数据库
MySQL数据库准备
00_mysql.yaml
apiVersion: v1
kind: Service
metadata:
name: mysql-test
namespace: default
spec:
ports:
- port: 3306
name: mysql
clusterIP: None
selector:
app: mysql-dump
---
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: db
namespace: default
spec:
selector:
matchLabels:
app: mysql-dump
serviceName: "mysql-test"
template:
metadata:
labels:
app: mysql-dump
spec:
nodeName: worker01
containers:
- name: mysql
image: mysql:5.7
env:
- name: MYSQL_ROOT_PASSWORD
value: "abc123"
ports:
- containerPort: 3306
volumeMounts:
- mountPath: "/var/lib/mysql"
name: mysql-data
volumes:
- name: mysql-data
hostPath:
path: /opt/mysqldata
Cronjob控制器类型应用资源清单文件
05_cronjob.yaml
apiVersion: batch/v1
kind: CronJob
metadata:
name: mysql-dump
spec:
schedule: "*/1 * * * *"
jobTemplate:
spec:
template:
spec:
nodeName: worker02
containers:
- name: c1
image: mysql:5.7
command: ["/bin/sh","-c","mysqldump --host=mysql-test -uroot -pabc123 --databases mysql > /root/mysql`date +%Y%m%d%H%M`.sql"]
volumeMounts:
- name: mysql-data
mountPath: "/root"
restartPolicy: Never
volumes:
- name: mysql-data
hostPath:
path: /opt/mysqldump
kubectl apply -f 00_mysql.yaml
kubectl apply -f 05_cronjob.yaml
# 查看验证
kubectl get cronjob
NAME SCHEDULE SUSPEND ACTIVE LAST SCHEDULE AGE
mysql-dump */1 * * * * False 0 30s 49s
# 看AGE时间,每分钟整点执行一次
kubectl get pods
NAME READY STATUS RESTARTS AGE
db-0 1/1 Running 0 4m
mysql-dump-28322104-fdmpg 0/1 Completed 0 2m33s
mysql-dump-28322105-dwc8r 0/1 Completed 0 93s
mysql-dump-28322106-cfxpl 0/1 Completed 0 33s
# worker02
ll /opt/mysqldump
total 11052
-rw-r--r-- 1 root root 3771736 Nov 7 11:04 mysql202311070304.sql
-rw-r--r-- 1 root root 3771736 Nov 7 11:05 mysql202311070305.sql
-rw-r--r-- 1 root root 3771736 Nov 7 11:06 mysql202311070306.sql
# 删除
kubectl delete -f 05_cronjob.yaml
kubectl delete -f 00_mysql.yaml
Controller之StatefulSet控制器
StatefulSet控制器作用
- StatefulSet 是用来管理有状态应用的控制器
- StatefulSet 用来管理某Pod集合的部署和扩缩， 并为这些 Pod 提供持久存储和持久标识符
- 参考: kubernetes.io/zh/docs/con…
无状态应用与有状态应用
无状态应用
- 如nginx
- 请求本身包含了响应端为响应这一请求所需的全部信息。每一个请求都像首次执行一样，不会依赖之前的数据进行响应
- 不需要持久化的数据
- 无状态应用的多个实例之间互不依赖，可以无序的部署、删除或伸缩
有状态应用
- 如mysql
- 前后请求有关联与依赖
- 需要持久化的数据
- 有状态应用的多个实例之间有依赖，不能相互替换：无论怎么调度，每个 Pod 都有一个永久不变的 ID
StatefulSet的特点
- 稳定的、唯一的网络标识符。(通过headless服务实现)
- 稳定的、持久的存储。(通过PV，PVC，storageclass实现)
- 有序的、优雅的部署和缩放
- 有序的、自动的滚动更新
StatefulSet的YAML组成
- headless service: 实现稳定，唯一的网络标识
- statefulset类型资源: 写法和deployment几乎一致，就是类型不一样
- volumeClaimTemplate : 指定存储卷
创建StatefulSet应用
参考: kubernetes.io/zh/docs/tut…
编辑YAML资源清单文件
创建statelfulset应用来调用名为nfs-client的storageclass，以实现动态供给，这里需要先参考
存储的动态供给创建storageclass
nginx-storageclass-nfs.yml
apiVersion: v1
kind: Service
metadata:
name: nginx
labels:
app: nginx
spec:
ports:
- port: 80
name: web
clusterIP: None # 无头服务
selector:
app: nginx
---
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: web # statefulset的名称
spec:
serviceName: "nginx"
replicas: 3 # 3个副本
selector:
matchLabels:
app: nginx
template:
metadata:
labels:
app: nginx
spec:
containers:
- name: nginx
image: nginx:1.15-alpine
ports:
- containerPort: 80
name: web
volumeMounts:
- name: www
mountPath: /usr/share/nginx/html
volumeClaimTemplates:
- metadata:
name: www
spec:
accessModes: [ "ReadWriteOnce" ]
storageClassName: "nfs-client" # 与前面定义的storageclass名称对应
resources:
requests:
storage: 1Gi
kubectl apply -f nginx-storageclass-nfs.yml
应用部署后验证
# 验证pod
kubectl get pods
NAME READY STATUS RESTARTS AGE
nfs-client-provisioner-5bffc5866c-h79v4 1/1 Running 0 86s
web-0 1/1 Running 0 36s
web-1 1/1 Running 0 34s
web-2 1/1 Running 0 31s
# 验证pv，自动产生了3个pv
kubectl get pv
NAME CAPACITY ACCESS MODES RECLAIM POLICY STATUS CLAIM STORAGECLASS REASON AGE
pvc-8f4dfaf6-e0bf-4681-8558-6be31c05e0e7 1Gi RWO Delete Bound default/www-web-2 nfs-client 50s
pvc-940a950e-e09e-4a71-a773-14c893e0014d 1Gi RWO Delete Bound default/www-web-1 nfs-client 53s
pvc-e31799a0-9bd3-4da8-bfd7-9f0ef74f7ade 1Gi RWO Delete Bound default/www-web-0 nfs-client 55s
# 验证pvc，自动产生了3个PVC
kubectl get pvc
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
www-web-0 Bound pvc-e31799a0-9bd3-4da8-bfd7-9f0ef74f7ade 1Gi RWO nfs-client 75s
www-web-1 Bound pvc-940a950e-e09e-4a71-a773-14c893e0014d 1Gi RWO nfs-client 73s
www-web-2 Bound pvc-8f4dfaf6-e0bf-4681-8558-6be31c05e0e7 1Gi RWO nfs-client 70s
# nfs 节点
# 在nfs服务器的共享目录中发现自动产生了3个子目录
ls /data/nfs/
default-www-web-0-pvc-e31799a0-9bd3-4da8-bfd7-9f0ef74f7ade default-www-web-1-pvc-940a950e-e09e-4a71-a773-14c893e0014d default-www-web-2-pvc-8f4dfaf6-e0bf-4681-8558-6be31c05e0e7
# master01
# 验证存储持久性
# 在web-0 pod中创建一个主页文件
kubectl exec -it web-0 -- /bin/sh -c "echo haha > /usr/share/nginx/html/index.html"
# nfs 节点
# 在nfs服务器上发现文件被创建到了对应的目录中
ls /data/nfs/*/
/data/nfs/default-www-web-0-pvc-e31799a0-9bd3-4da8-bfd7-9f0ef74f7ade/:
index.html
/data/nfs/default-www-web-1-pvc-940a950e-e09e-4a71-a773-14c893e0014d/:
/data/nfs/default-www-web-2-pvc-8f4dfaf6-e0bf-4681-8558-6be31c05e0e7/:
# 文件内的内容也与web-0的pod中创建的一致
cat /data/nfs/default-www-web-0-pvc-e31799a0-9bd3-4da8-bfd7-9f0ef74f7ade/index.html
haha
# master01
# 删除web-0这个pod，再验证
kubectl delete pod web-0
# 因为控制器的原因，会迅速再拉起web-0这个pod
kubectl get pods
NAME READY STATUS RESTARTS AGE
nfs-client-provisioner-5bffc5866c-h79v4 1/1 Running 0 10m
web-0 1/1 Running 0 5s
web-1 1/1 Running 0 9m47s
web-2 1/1 Running 0 9m44s
# 新拉起的pod仍然是相同的存储数据
kubectl exec -it web-0 -- cat /usr/share/nginx/html/index.html
haha
访问验证
# 验证Coredns是否可用
kubectl get svc -n kube-system
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
kube-dns ClusterIP 10.96.0.10 <none> 53/UDP,53/TCP,9153/TCP 12d
dig -t a www.baidu.com @10.96.0.10
...
;; ANSWER SECTION:
www.baidu.com. 30 IN CNAME www.a.shifen.com.
www.a.shifen.com. 30 IN A 183.2.172.185
www.a.shifen.com. 30 IN A 183.2.172.42
;; Query time: 2020 msec
;; SERVER: 10.96.0.10#53(10.96.0.10)
;; WHEN: Tue Nov 14 16:22:22 CST 2023
;; MSG SIZE rcvd: 149
dig -t a nginx.default.svc.cluster.local. @10.96.0.10
...
;; ANSWER SECTION:
nginx.default.svc.cluster.local. 30 IN A 10.244.5.21
nginx.default.svc.cluster.local. 30 IN A 10.244.5.23
nginx.default.svc.cluster.local. 30 IN A 10.244.30.84
;; Query time: 13 msec
;; SERVER: 10.96.0.10#53(10.96.0.10)
;; WHEN: Tue Nov 14 16:23:02 CST 2023
;; MSG SIZE rcvd: 201
dig -t a web-0.nginx.default.svc.cluster.local. @10.96.0.10
...
;; ANSWER SECTION:
web-0.nginx.default.svc.cluster.local. 30 IN A 10.244.30.84
;; Query time: 5 msec
;; SERVER: 10.96.0.10#53(10.96.0.10)
;; WHEN: Tue Nov 14 16:24:10 CST 2023
;; MSG SIZE rcvd: 119
# 在kubernetes集群内创建pod访问
# 创建容器，并进入容器
kubectl run -i --tty --rm=true busybox --image=radial/busyboxplus
# 该请求会路由到web-0，web-1，web-2中的一个
curl nginx.default.svc.cluster.local.
haha
curl nginx.default.svc.cluster.local.
<html>
<head><title>403 Forbidden</title></head>
curl web-0.nginx.default.svc.cluster.local.
haha
exit
已部署应用滚动更新(含金丝雀发布)
它将按照与 Pod 终止相同的顺序（从最大序号到最小序号）进行，每次更新一个 Pod。StatefulSet可以使用partition参数来实现金丝雀更新，partition参数可以控制StatefulSet控制器更新的Pod。下面，我们就进行StatefulSet控制器的金丝雀更新实战
# 使用patch参数来指定了StatefulSet控制器的partition参数为2，表示当更新时，只有Pod的编号大于等于2的才更新。
kubectl patch sts web -p '{"spec":{"updateStrategy":{"rollingUpdate":{"partition":2}}}}'
kubectl exec -it web-0 -- nginx -v
nginx version: nginx/1.15.12
kubectl exec -it web-1 -- nginx -v
nginx version: nginx/1.15.12
kubectl exec -it web-2 -- nginx -v
nginx version: nginx/1.15.12
kubectl set image sts/web nginx=nginx:latest
kubectl exec -it web-0 -- nginx -v
nginx version: nginx/1.15.12
kubectl exec -it web-1 -- nginx -v
nginx version: nginx/1.15.12
kubectl exec -it web-2 -- nginx -v
nginx version: nginx/1.25.3
kubectl get pods -o custom-columns=Name:metadata.name,Image:spec.containers[0].image
Name Image
nfs-client-provisioner-5bffc5866c-h79v4 registry.cn-beijing.aliyuncs.com/pylixm/nfs-subdir-external-provisioner:v4.0.0
web-0 nginx:1.15-alpine
web-1 nginx:1.15-alpine
web-2 nginx:latest
# 如何实现全部更新呢？
kubectl patch sts web -p '{"spec":{"updateStrategy":{"rollingUpdate":{"partition":0}}}}'
kubectl set image sts/web nginx=nginx:latest
kubectl get pods -o custom-columns=Name:metadata.name,Image:spec.containers[0].image
Name Image
nfs-client-provisioner-5bffc5866c-h79v4 registry.cn-beijing.aliyuncs.com/pylixm/nfs-subdir-external-provisioner:v4.0.0
web-0 nginx:latest
web-1 nginx:latest
web-2 nginx:latest
已部署应用扩容与缩容
在StatefulSet扩容时，会创建一个新的Pod，该Pod与之前的所有Pod都是有顺序的，并且新Pod的序号最大。在缩容时，StatefulSet控制器删除的也是序号最大的Pod
kubectl scale sts web --replicas=4
kubectl get pods
NAME READY STATUS RESTARTS AGE
nfs-client-provisioner-5bffc5866c-h79v4 1/1 Running 0 37m
web-0 1/1 Running 0 101s
web-1 1/1 Running 0 103s
web-2 1/1 Running 0 5m24s
web-3 1/1 Running 0 18s
kubectl scale sts web --replicas=2
kubectl get pods
NAME READY STATUS RESTARTS AGE
nfs-client-provisioner-5bffc5866c-h79v4 1/1 Running 0 37m
web-0 1/1 Running 0 2m9s
web-1 1/1 Running 0 2m11s
# 删除
kubectl delete -f nginx-storageclass-nfs.yml
rm -f nginx-storageclass-nfs.yml
