kubernetes 之 Rolling Update 滚动升级

滚动升级

1、错误的yml文件

[machangwei@mcwk8s-master ~]$ cat mcwHttpd.yml

apiVersion: apps/v1

kind: Deployment

metadata:

  name: mcw-httpd

spec:

  replicas: 3

  selector:

    matchLabels:

     run: mcw-httpd

  template:

    metadata:

      labels:

        run: mcw-httpd

    spec:

      containers:

        - name: mcw-httpd

          image: httpd

          ports:

          - containerPort: 80

[machangwei@mcwk8s-master ~]$ vim mcwHttpd.yml

[machangwei@mcwk8s-master ~]$ cat mcwHttpd.yml

apiVersion: apps/v1

kind: Deployment

metadata:

  name: mcw-httpd

spec:

  replicas: 3

  selector:

    matchLabels:

     run: mcw-httpd

  template:

    metadata:

      labels:

        run: mcw-httpd

    spec:

      containers:

        - name: mcw-httpd

          image: :httpd:2.2.31

          ports:

          - containerPort: 80

[machangwei@mcwk8s-master ~]$ kubectl apply -f mcwHttpd.yml

deployment.apps/mcw-httpd created

[machangwei@mcwk8s-master ~]$ kubectl get deployment httpd -o wide

Error from server (NotFound): deployments.apps "httpd" not found

[machangwei@mcwk8s-master ~]$ kubectl get deployment -o wide

NAME        READY   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS   IMAGES          SELECTOR

mcw-httpd   0/3     3            0           33s   mcw-httpd    :httpd:2.2.31   run=mcw-httpd

[machangwei@mcwk8s-master ~]$ kubectl get pod

NAME                         READY   STATUS             RESTARTS   AGE

mcw-httpd-6644b55f8c-62dvc   0/1     InvalidImageName   0          98s

mcw-httpd-6644b55f8c-h8pfn   0/1     InvalidImageName   0          98s

mcw-httpd-6644b55f8c-xn259   0/1     InvalidImageName   0          98s

[machangwei@mcwk8s-master ~]$

2、修改为正确的yml文件，部署deployment

[machangwei@mcwk8s-master ~]$ vim mcwHttpd.yml

[machangwei@mcwk8s-master ~]$ cat mcwHttpd.yml  #其它跟以前的没变，只是镜像后面加了冒号和数字，也就是添加了版本

apiVersion: apps/v1

kind: Deployment

metadata:

  name: mcw-httpd

spec:

  replicas: 3

  selector:

    matchLabels:

     run: mcw-httpd

  template:

    metadata:

      labels:

        run: mcw-httpd

    spec:

      containers:

        - name: mcw-httpd

          image: httpd:2.2.31

          ports:

          - containerPort: 80

[machangwei@mcwk8s-master ~]$ kubectl delete -f mcwHttpd.yml #把以前的删除

deployment.apps "mcw-httpd" deleted

[machangwei@mcwk8s-master ~]$

[machangwei@mcwk8s-master ~]$ kubectl apply -f mcwHttpd.yml #部署deployment

deployment.apps/mcw-httpd created

[machangwei@mcwk8s-master ~]$ kubectl get deployment mcw-httpd -o wide  #查看镜像是31的镜像

NAME        READY   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS   IMAGES         SELECTOR

mcw-httpd   3/3     3            3           13s   mcw-httpd    httpd:2.2.31   run=mcw-httpd

[machangwei@mcwk8s-master ~]$ kubectl get replicaset  -o wide #查看副本集这里，也是31的镜像

NAME                   DESIRED   CURRENT   READY   AGE   CONTAINERS   IMAGES         SELECTOR

mcw-httpd-6f64bbcfcf   3         3         3       40s   mcw-httpd    httpd:2.2.31   pod-template-hash=6f64bbcfcf,run=mcw-httpd

[machangwei@mcwk8s-master ~]$ kubectl get pod -o wide  #查看pod部署情况

NAME                         READY   STATUS    RESTARTS   AGE   IP           NODE           NOMINATED NODE   READINESS GATES

mcw-httpd-6f64bbcfcf-5nhxc   1/1     Running   0          70s   10.244.2.9   mcwk8s-node2   <none>           <none>

mcw-httpd-6f64bbcfcf-s8hng   1/1     Running   0          70s   10.244.1.7   mcwk8s-node1   <none>           <none>

mcw-httpd-6f64bbcfcf-t6j2f   1/1     Running   0          70s   10.244.2.8   mcwk8s-node2   <none>           <none>

3、修改镜像版本号，重新部署，滚动升级

[machangwei@mcwk8s-master ~]$ vim mcwHttpd.yml

[machangwei@mcwk8s-master ~]$ cat mcwHttpd.yml #修改版本号为：2.2.32

apiVersion: apps/v1

kind: Deployment

metadata:

  name: mcw-httpd

spec:

  replicas: 3

  selector:

    matchLabels:

     run: mcw-httpd

  template:

    metadata:

      labels:

        run: mcw-httpd

    spec:

      containers:

        - name: mcw-httpd

          image: httpd:2.2.32

          ports:

          - containerPort: 80

[machangwei@mcwk8s-master ~]$ kubectl apply -f mcwHttpd.yml #重新部署

deployment.apps/mcw-httpd configured

[machangwei@mcwk8s-master ~]$ kubectl get deployment mcw-httpd -o wide  #查看deployment镜像已经修改，更新了一个好像是

NAME        READY   UP-TO-DATE   AVAILABLE   AGE     CONTAINERS   IMAGES         SELECTOR

mcw-httpd   3/3     1            3           6m58s   mcw-httpd    httpd:2.2.32   run=mcw-httpd

[machangwei@mcwk8s-master ~]$ kubectl get replicaset  -o wide #查看副本集，32是新建的副本集，镜像改为32了。这里还没有升级完，32镜像只有一个是ready，也就是刚刚升级完一个

NAME                   DESIRED   CURRENT   READY   AGE    CONTAINERS   IMAGES         SELECTOR

mcw-httpd-6f64bbcfcf   2         2         2       7m5s   mcw-httpd    httpd:2.2.31   pod-template-hash=6f64bbcfcf,run=mcw-httpd

mcw-httpd-7c977464db   2         2         1       10s    mcw-httpd    httpd:2.2.32   pod-template-hash=7c977464db,run=mcw-httpd

[machangwei@mcwk8s-master ~]$

[machangwei@mcwk8s-master ~]$ kubectl get pod -o wide #查看pod。根据年龄查看，刚刚生成三个新的，旧的还有一个没有被删除

NAME                         READY   STATUS        RESTARTS   AGE     IP            NODE           NOMINATED NODE   READINESS GATES

mcw-httpd-6f64bbcfcf-s8hng   1/1     Terminating   0          7m16s   10.244.1.7    mcwk8s-node1   <none>           <none>

mcw-httpd-7c977464db-5ctkr   1/1     Running       0          21s     10.244.1.8    mcwk8s-node1   <none>           <none>

mcw-httpd-7c977464db-bfz6t   1/1     Running       0          10s     10.244.1.9    mcwk8s-node1   <none>           <none>

mcw-httpd-7c977464db-gp94c   1/1     Running       0          12s     10.244.2.10   mcwk8s-node2   <none>           <none>

[machangwei@mcwk8s-master ~]$

[machangwei@mcwk8s-master ~]$ kubectl get deployment mcw-httpd -o wide  #再次查看，已经升级完三个了

NAME        READY   UP-TO-DATE   AVAILABLE   AGE     CONTAINERS   IMAGES         SELECTOR

mcw-httpd   3/3     3            3           7m36s   mcw-httpd    httpd:2.2.32   run=mcw-httpd

[machangwei@mcwk8s-master ~]$ kubectl get replicaset  -o wide  #再次查看副本集，三个副本都已经被新生成副本代替了，

#新生成的副本是 32的镜像。而且旧的副本集副本个数为0了，但是它不会删除过了九分钟都还没有删除，应该是不删的

NAME                   DESIRED   CURRENT   READY   AGE     CONTAINERS   IMAGES         SELECTOR

mcw-httpd-6f64bbcfcf   0         0         0       7m47s   mcw-httpd    httpd:2.2.31   pod-template-hash=6f64bbcfcf,run=mcw-httpd

mcw-httpd-7c977464db   3         3         3       52s     mcw-httpd    httpd:2.2.32   pod-template-hash=7c977464db,run=mcw-httpd

[machangwei@mcwk8s-master ~]$ kubectl get pod -o wide #查看pod，已经只有最新生成的镜像是32版本的pod了

NAME                         READY   STATUS    RESTARTS   AGE   IP            NODE           NOMINATED NODE   READINESS GATES

mcw-httpd-7c977464db-5ctkr   1/1     Running   0          57s   10.244.1.8    mcwk8s-node1   <none>           <none>

mcw-httpd-7c977464db-bfz6t   1/1     Running   0          46s   10.244.1.9    mcwk8s-node1   <none>           <none>

mcw-httpd-7c977464db-gp94c   1/1     Running   0          48s   10.244.2.10   mcwk8s-node2   <none>           <none>

4、查看滚动升级详情

[machangwei@mcwk8s-master ~]$ kubectl get replicaset  -o wide #查看副本集

NAME                   DESIRED   CURRENT   READY   AGE   CONTAINERS   IMAGES         SELECTOR

mcw-httpd-6f64bbcfcf   0         0         0       21m   mcw-httpd    httpd:2.2.31   pod-template-hash=6f64bbcfcf,run=mcw-httpd

mcw-httpd-7c977464db   3         3         3       14m   mcw-httpd    httpd:2.2.32   pod-template-hash=7c977

[machangwei@mcwk8s-master ~]$ kubectl get deployment -o wide

NAME        READY   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS   IMAGES         SELECTOR

mcw-httpd   3/3     3            3           19m   mcw-httpd    httpd:2.2.32   run=mcw-httpd

[machangwei@mcwk8s-master ~]$ kubectl describe deployment mcw-httpd

Replicas:               3 desired | 3 updated | 3 total | 3 available | 0 unavailable #副本三个被更新的

 Image:        httpd:2.2.32  容器镜像是32

Events:

  Type    Reason             Age   From                   Message

  ----    ------             ----  ----                   -------

  Normal  ScalingReplicaSet  19m   deployment-controller  Scaled up replica set mcw-httpd-6f64bbcfcf to 3

  Normal  ScalingReplicaSet  12m   deployment-controller  Scaled up replica set mcw-httpd-7c977464db to 1

  Normal  ScalingReplicaSet  12m   deployment-controller  Scaled down replica set mcw-httpd-6f64bbcfcf to 2

  Normal  ScalingReplicaSet  12m   deployment-controller  Scaled up replica set mcw-httpd-7c977464db to 2

  Normal  ScalingReplicaSet  12m   deployment-controller  Scaled down replica set mcw-httpd-6f64bbcfcf to 1

  Normal  ScalingReplicaSet  12m   deployment-controller  Scaled up replica set mcw-httpd-7c977464db to 3

  Normal  ScalingReplicaSet  12m   deployment-controller  Scaled down replica set mcw-httpd-6f64bbcfcf to 0

 每只更新替换一个pod：

mcw-httpd-6f64bbcfcf to 3  旧副本集是3个pod

mcw-httpd-7c977464db to 1  新副本集增加一个pod，总数为1

mcw-httpd-6f64bbcfcf to 2  旧副本集减少一个pod，总数为2

mcw-httpd-7c977464db to 2  新副本集增加一个pod，总数为2

mcw-httpd-6f64bbcfcf to 1  旧副本集减少一个pod，总数为1

mcw-httpd-7c977464db to 3  新副本集增加一个pod，总数为3

mcw-httpd-6f64bbcfcf to 0  旧副本集减少一个pod，总数为0

每次替换的POD数量是可以定制的。默认应该是1吧。定制的参数有两个，如：maxSurge ,maxUnavailable 来精细控制POD的替换数量

回滚

1、回滚前的模拟准备环境

模拟新建三个需要的版本。

[root@mcwk8s-node2 ~]$ docker tag httpd:latest  httpd:2.4.16

[root@mcwk8s-node2 ~]$ docker tag httpd:latest  httpd:2.4.17

[root@mcwk8s-node2 ~]$ docker tag httpd:latest  httpd:2.4.18

创文件文件，并修改版本

[machangwei@mcwk8s-master ~]$ cp mcwHttpd.yml mcw.httpd.v1.yml

[machangwei@mcwk8s-master ~]$ cp mcwHttpd.yml mcw.httpd.v2.yml

[machangwei@mcwk8s-master ~]$ cp mcwHttpd.yml mcw.httpd.v3.yml

修改一下文件，将镜像分别修改为  16,17,18

[machangwei@mcwk8s-master ~]$ ls mcw.httpd.v1

mcw.httpd.v16.yml  mcw.httpd.v17.yml  mcw.httpd.v18.yml

[machangwei@mcwk8s-master ~]$ cat mcw.httpd.v16.yml

apiVersion: apps/v1

kind: Deployment

metadata:

  name: mcw-httpd

spec:

  replicas: 3

  selector:

    matchLabels:

     run: mcw-httpd

  template:

    metadata:

      labels:

        run: mcw-httpd

    spec:

      containers:

        - name: mcw-httpd

          image: httpd:2.4.16

          ports:

          - containerPort: 80

2、部署并回滚

[machangwei@mcwk8s-master ~]$ ls mcw.httpd.v1 #查看所有yml配置文件

mcw.httpd.v16.yml  mcw.httpd.v17.yml  mcw.httpd.v18.yml

[machangwei@mcwk8s-master ~]$ kubectl apply -f mcw.httpd.v16.yml --record   #--record，将当前命令记录到revision记录中

Flag --record has been deprecated, --record will be removed in the future  #这样我们就能知道每个revison对应的是哪个配置文件了

deployment.apps/mcw-httpd created

[machangwei@mcwk8s-master ~]$

[machangwei@mcwk8s-master ~]$ kubectl get deployment mcw-httpd -o wide

NAME        READY   UP-TO-DATE   AVAILABLE   AGE    CONTAINERS   IMAGES         SELECTOR

mcw-httpd   3/3     3            3           2m3s   mcw-httpd    httpd:2.4.16   run=mcw-httpd

[machangwei@mcwk8s-master ~]$

[machangwei@mcwk8s-master ~]$ kubectl apply -f mcw.httpd.v17.yml --record

Flag --record has been deprecated, --record will be removed in the future

deployment.apps/mcw-httpd configured

[machangwei@mcwk8s-master ~]$ kubectl get deployment mcw-httpd -o wide  #每次执行都被更新镜像版本了

NAME        READY   UP-TO-DATE   AVAILABLE   AGE     CONTAINERS   IMAGES         SELECTOR

mcw-httpd   3/3     3            3           2m35s   mcw-httpd    httpd:2.4.17   run=mcw-httpd

[machangwei@mcwk8s-master ~]$

[machangwei@mcwk8s-master ~]$ kubectl apply -f mcw.httpd.v18.yml --record

Flag --record has been deprecated, --record will be removed in the future

deployment.apps/mcw-httpd configured

[machangwei@mcwk8s-master ~]$ kubectl get deployment mcw-httpd -o wide

NAME        READY   UP-TO-DATE   AVAILABLE   AGE     CONTAINERS   IMAGES         SELECTOR

mcw-httpd   3/3     2            3           2m54s   mcw-httpd    httpd:2.4.18   run=mcw-httpd

[machangwei@mcwk8s-master ~]$

[machangwei@mcwk8s-master ~]$ kubectl rollout history deployment mcw-httpd #--record，将当前命令记录到revision记录中

deployment.apps/mcw-httpd

REVISION  CHANGE-CAUSE    #这样我们就能知道每个revison对应的是哪个配置文件了  #查看revison历史记录

1         kubectl apply --filename=mcw.httpd.v16.yml --record=true

2         kubectl apply --filename=mcw.httpd.v17.yml --record=true

3         kubectl apply --filename=mcw.httpd.v18.yml --record=true

[machangwei@mcwk8s-master ~]$

[machangwei@mcwk8s-master ~]$ kubectl rollout undo deployment mcw-httpd --to-revision=1 #执行命令回滚到版本1。

deployment.apps/mcw-httpd rolled back

[machangwei@mcwk8s-master ~]$

[machangwei@mcwk8s-master ~]$ kubectl get deployment mcw-httpd -o wide #查看镜像已经改成版本1的了

NAME        READY   UP-TO-DATE   AVAILABLE   AGE     CONTAINERS   IMAGES         SELECTOR

mcw-httpd   3/3     3            3           4m39s   mcw-httpd    httpd:2.4.16   run=mcw-httpd

[machangwei@mcwk8s-master ~]$

[machangwei@mcwk8s-master ~]$ kubectl rollout history deployment mcw-httpd #查看revison历史记录的变化

deployment.apps/mcw-httpd

REVISION  CHANGE-CAUSE

2         kubectl apply --filename=mcw.httpd.v17.yml --record=true

3         kubectl apply --filename=mcw.httpd.v18.yml --record=true

4         kubectl apply --filename=mcw.httpd.v16.yml --record=true #revison 1 变成了revison4，我们可以通过CHANG-CAUSE

#知道每个revison的具体含义，所以要在执行kubectl 加上参数 --record

[machangwei@mcwk8s-master ~]$

kubernetes 之 Rolling Update 滚动升级的更多相关文章

使用 kubectl 执行 Rolling Update（滚动更新）
Rolling Update滚动更新通过使用新版本的 Pod 逐步替代旧版本的 Pod 来实现 Deployment 的更新,从而实现零停机.新的 Pod 将在具有可用资源的 Node(节点)上进行 ...
kubernetes云平台管理实战：滚动升级秒级回滚（六）
一.nginx保证有两个版本 1.查看当前容器运行nginx版本 [root@k8s-master ~]# kubectl get pod -o wide NAME READY STATUS REST ...
Kubernetes Deployment滚动升级
我们k8s集群使用的是1.7.7版本的,该版本中官方已经推荐使用Deployment代替Replication Controller(rc)了,Deployment继承了rc的全部功能外,还可以查看升 ...
入门Kubernetes - 滚动升级/回滚
一.前言上一篇文章中对yaml文件格式进行了解,并对k8s中各种主要资源通过yaml创建时的定义模板.接来下就进一步学习k8s的各种特点.并应用在示例中. 接下来先实现.Net Core Api程序 ...
kubernetes之deployment滚动升级
参考:https://blog.51cto.com/wutengfei/2116663 创建和管理多个Pod--Deployment Deployment 为 Pod 和 ReplicaSet 提供了 ...
k8s滚动升级
为了服务升级过程中提供可持续的不中断的服务,Kubernetes 提供了rolling update机制,具体配置需要修改对应服务的yaml文件参数解析: minReadySeconds: 100 ...
ASP.NET Core on K8S深入学习（5）Rolling Update
本篇已加入<.NET Core on K8S学习实践系列文章索引>,可以点击查看更多容器化技术相关系列文章. 一.什么是Rolling Update? 为了服务升级过程中提供可持续的不中断 ...
K8S服务滚动升级
对于Kubernetes集群来说,一个service可能有多个pod,滚动升级(Rolling update)就是指每次更新部分Pod,而不是在同一时刻将该Service下面的所有Pod shutdo ...
在 Rolling Update 中使用 Health Check - 每天5分钟玩转 Docker 容器技术（146）
上一节讨论了 Health Check 在 Scale Up 中的应用,Health Check 另一个重要的应用场景是 Rolling Update.试想一下下面的情况: 现有一个正常运行的多副本应 ...
在 Rolling Update 中使用 Health Check【转】
上一节讨论了 Health Check 在 Scale Up 中的应用,Health Check 另一个重要的应用场景是 Rolling Update.试想一下下面的情况: 现有一个正常运行的多副本应 ...

随机推荐

reactive stream协议详解
目录背景什么是reactive stream 深入了解java版本的reactive stream Publisher Subscriber Subscription Processor JDK中 ...
在linux上使用Qt开发动态库项目，怎么只生成一个so文件
背景: 在linux系统上,我们使用 Qt 开发动态库项目时,会默认生成四个文件:x.so .x.so.1 .x.so.1.0.x.so.1.0.0 四个文件,只有一个真实的so库,剩下的三个都是链 ...
Python操作临时文件---tempfile
# 使用标准库中 tempfile 下的 TemporaryFile,NamedTemporaryFile # TemporaryFile(mode='w+b', bufsize=1, suffix= ...
高可用之战：Redis Sentinal(哨兵模式）
★ Redis24篇集合 1 背景在我们的<Redis高可用之战:主从架构>篇章中,介绍了Redis的主从架构模式,可以有效的提升Redis服务的可用性,减少甚至避免Redis服务发生完 ...
国产开源数据库OpenGauss的安装运行
步骤一:OpenGauss 的安装环境 OS:openEuler 20.03 64bit with ARM 架构:arm64 部署:单机安装过程 1.环境配置安装依赖包: yum install ...
什么是coredump
什么是 coredump 介绍在 Linux 开发中,我们经常听到程序员说我的程序 core 掉了,通常出现这类的问题是低级 bug 中的内存访问越界.使用空指针.堆栈溢出等情况.使程序运行过程中异 ...
选择排序的基本实现【数据结构与算法—TypeScript 实现】
笔记整理自 coderwhy 『TypeScript 高阶数据结构与算法』课程概念本质:两两元素相比较,先扫描一遍未排序数列,把未排序的数列中的最小(大)元素,放到数列的已排序的末尾特性选择排 ...
HarmonyOS API Version 7版本特性说明
2020年9月11日,HarmonyOS SDK发布了首个Beta版本,支持基于HarmonyOS的华为智慧屏.智能穿戴.车机设备开发,让广大的开发者正式步入了HarmonyOS应用开发之旅. 开发者 ...
Java入门01：环境安装与配置
Java入门01:环境安装与配置 JDK下载下载地址:https://www.oracle.com/cn/java/technologies/javase/javase-jdk8-downloads ...
Flask搭建APP统一管理平台
主页效果: 1.从数据库中获取所有APP的信息,每个卡片上展示APP名称.bundle id.版本构建历史记录,系统类型等构建信息 2.支持查询筛选,模糊查询 3.点击历史记录跳转APP历史记录详情页 ...

kubernetes 之 Rolling Update 滚动升级

滚动升级

1、错误的yml文件

2、修改为正确的yml文件，部署deployment

3、修改镜像版本号，重新部署，滚动升级

4、查看滚动升级详情

回滚

1、回滚前的模拟准备环境

2、部署并回滚

kubernetes 之 Rolling Update 滚动升级的更多相关文章

随机推荐

热门专题