一、知识准备
● 本文详细探索deployment在滚动更新时候的行为
二、环境准备
| OS |
Ubuntu 18.04.1 LTS |
| docker |
18.06.0-ce |
三、准备镜像
首先准备2个不同版本的镜像，用于测试（已经在阿里云上创建好2个不同版本的nginx镜像）
docker pull registry.cn-beijing.aliyuncs.com/mrvolleyball/nginx:v1docker pull registry.cn-beijing.aliyuncs.com/mrvolleyball/nginx:v2
root@k8s-master:~# docker run -d --rm -p 10080:80 nginx:v1e88097841c5feef92e4285a2448b943934ade5d86412946bc8d86e262f80a050root@k8s-master:~# curl http://127.0.0.1:10080----------version: v1hostname: f5189a5d3ad3
四、deployment、replicaset、pod之间的关系
+------------+ | deployment | +-----+------+ | | | | +--------------------------------------------------+ | | | | | | | | | | | | | | | | | |+------v------+ +------v------+ +------v------+|replicaset:v1| |replicaset:v2| |replicaset:v3|+-------------+ +------+------+ +-------------+ | | +--------+---------+ | | | | +---v---+ +---v---+ |pod:v2 | |pod:v2 | +-------+ +-------+
● deployment调度replicaset，pod由replicaset调度
● deployment管理多个replicaset版本，可用于回滚
● replicaset控制pod的行为，包括新增pod、删除pod
我们首先准备一个yaml文件用于测试：
root@k8s-master:~# more roll_update.yamlapiVersion: extensions/v1beta1kind: Deploymentmetadata: name: image-deploymentspec: replicas: 1 template: metadata: labels: app: image-update spec: containers: - name: nginx image: registry.cn-beijing.aliyuncs.com/mrvolleyball/nginx:v1 imagePullPolicy: Always
简单验证一下：
root@k8s-master:~# kubectl apply -f roll_update.yamldeployment.extensions "update-deployment" created
root@k8s-master:~# kubectl get deployNAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGEupdate-deployment 3 3 3 3 54sroot@k8s-master:~# kubectl get rsNAME DESIRED CURRENT READY AGEupdate-deployment-7db77f7cc6 3 3 3 56sroot@k8s-master:~# kubectl get podNAME READY STATUS RESTARTS AGEupdate-deployment-7db77f7cc6-7j49g 1/1 Running 0 1mupdate-deployment-7db77f7cc6-b75wn 1/1 Running 0 1mupdate-deployment-7db77f7cc6-cfnt5 1/1 Running 0 1m
deployment、replicaset、pod都已经正常启动，下面分析一下他们的行为：
deployment
root@k8s-master:~# kubectl describe deploy update-deploymentName: update-deploymentNamespace: default...Replicas: 3 desired | 3 updated | 3 total | 3 available | 0 unavailableStrategyType: RollingUpdateMinReadySeconds: 0RollingUpdateStrategy: 1 max unavailable, 1 max surge...NewReplicaSet: update-deployment-7db77f7cc6 (3/3 replicas created)Events: Type Reason Age From Message ---- ------ ---- ---- ------- Normal ScalingReplicaSet 1m deployment-controller Scaled up replica set update-deployment-7db77f7cc6 to 3
● deployment创建了一个replicaset，叫做update-deployment-7db77f7cc6（7db77f7cc6是replicaset的template hash值）
● 根据配置文件的要求，replicaset的副本数为3
replicaset
root@k8s-master:~# kubectl describe rs update-deployment-7db77f7cc6Name: update-deployment-7db77f7cc6Namespace: default...Controlled By: Deployment/update-deploymentReplicas: 3 current / 3 desiredPods Status: 3 Running / 0 Waiting / 0 Succeeded / 0 Failed...Events: Type Reason Age From Message ---- ------ ---- ---- ------- Normal SuccessfulCreate 3m replicaset-controller Created pod: update-deployment-7db77f7cc6-7j49g Normal SuccessfulCreate 3m replicaset-controller Created pod: update-deployment-7db77f7cc6-b75wn Normal SuccessfulCreate 3m replicaset-controller Created pod: update-deployment-7db77f7cc6-cfnt5
● replicaset创建了3个pod
pod
root@k8s-master:~# kubectl describe pod update-deployment-7db77f7cc6-7j49gName: update-deployment-7db77f7cc6-7j49gNamespace: default...Status: RunningIP: 10.10.169.140Controlled By: ReplicaSet/update-deployment-7db77f7cc6...Events: Type Reason Age From Message ---- ------ ---- ---- ------- Normal Scheduled 9m default-scheduler Successfully assigned update-deployment-7db77f7cc6-7j49g to k8s-node2 Normal SuccessfulMountVolume 9m kubelet, k8s-node2 MountVolume.SetUp succeeded for volume "default-token-v9nkm" Normal Pulling 9m kubelet, k8s-node2 pulling image "registry.cn-beijing.aliyuncs.com/mrvolleyball/nginx:v1" Normal Pulled 9m kubelet, k8s-node2 Successfully pulled image "registry.cn-beijing.aliyuncs.com/mrvolleyball/nginx:v1" Normal Created 9m kubelet, k8s-node2 Created container Normal Started 9m kubelet, k8s-node2 Started container
● pod被replicaset创建之后，开始分配到worker节点、拉取镜像、启动容器等一系列操作
● 所以pod的命名方式是：update-deployment-7db77f7cc6-7j49g（deployment名字-replicaset模板hash名字-pod模板hash名字）
不禁有同学要问，为什么搞这么复杂，启动一个pod需要动用这么多组件呢？下面用一个场景说明为啥需要这么多组件：
镜像版本更新
● 当镜像版本有更新时（三种方法都可以实现，参考前一篇文章：更新k8s镜像版本的三种方式），既要保证服务可用，又要保证在线更新，流程应该是：
??1、先增加一个pod，镜像版本为新版本
??2、pod可用之后，删除一个老版本pod
??3、循环第1、2步，直到老版本pod全部删除，新版本的pod全部可用
● 上述的这个过程就是replicaset的作用，它根据需求，自动的增加新版本pod，然后删除老版本pod，直到老版本pod全部删除，新版本的pod全部可用
● 如果此时版本需要回退，那replicaset需要把刚才的步骤逆向更新一遍，实现版本回退
● deployment的作用就是管理replicaset。deployment会保存各个版本的replicaset，一旦需要进行版本回滚，deployment会立即回滚replicaset的版本，从而控制pod状态
下面测试一下：
使用patch命令更新镜像版本，并且使用pause命令来观察：
root@k8s-master:~# kubectl patch deployment update-deployment --patch '{"spec": {"template": {"spec": {"containers": [{"name": "nginx","image":"registry.cn-beijing.aliyuncs.com/mrvolleyball/nginx:v2"}]}}}}' && kubectl rollout pause deployment update-deploymentdeployment.extensions "update-deployment" patcheddeployment.apps "update-deployment" paused
此时pod状态：
root@k8s-master:~# kubectl get pod -owideNAME READY STATUS RESTARTS AGE IP NODEupdate-deployment-7db77f7cc6-7j49g 1/1 Running 0 1h 10.10.169.140 k8s-node2update-deployment-7db77f7cc6-b75wn 1/1 Running 0 1h 10.10.235.211 k8s-masterupdate-deployment-7db77f7cc6-cfnt5 1/1 Terminating 0 1h 10.10.36.126 k8s-node1update-deployment-7fb7b4b557-6987x 1/1 Running 0 7s 10.10.36.127 k8s-node1update-deployment-7fb7b4b557-dxdqb 1/1 Running 0 10s 10.10.169.139 k8s-node2
新增了2个pod，而删除了1个老版本的pod
此时replicaset状态：
root@k8s-master:~# kubectl get rs -owideNAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTORupdate-deployment-7db77f7cc6 2 2 2 1h nginx registry.cn-beijing.aliyuncs.com/mrvolleyball/nginx:v1 app=roll-update,pod-template-hash=3863393772update-deployment-7fb7b4b557 2 2 2 4m nginx registry.cn-beijing.aliyuncs.com/mrvolleyball/nginx:v2 app=roll-update,pod-template-hash=3963606113
有一个新版本的replicaset创建了出来，并且需求的pod数量为2，而原来的replicaset需求的pod数量从3降为2
查看replicaset版本：
root@k8s-master:~# kubectl rollout history deploy update-deploymentdeployments "update-deployment"REVISION CHANGE-CAUSE1 <none>2 update version to v2
新增了一个版本2
由于使用pause命令，更新过程到此会卡主，我们让更新的过程继续下去：
root@k8s-master:~# kubectl rollout resume deployment update-deploymentdeployment.apps "update-deployment" resumed
查看状态：
root@k8s-master:~# kubectl get podNAME READY STATUS RESTARTS AGEupdate-deployment-7fb7b4b557-6987x 1/1 Running 0 15mupdate-deployment-7fb7b4b557-dxdqb 1/1 Running 0 15mupdate-deployment-7fb7b4b557-wg5c8 1/1 Running 0 1mroot@k8s-master:~# kubectl get rs -owideNAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTORupdate-deployment-7db77f7cc6 0 0 0 1h nginx registry.cn-beijing.aliyuncs.com/mrvolleyball/nginx:v1 app=roll-update,pod-template-hash=3863393772update-deployment-7fb7b4b557 3 3 3 14m nginx registry.cn-beijing.aliyuncs.com/mrvolleyball/nginx:v2 app=roll-update,pod-template-hash=3963606113
v1版本的replicaset已经没有pod，但是历史记录还是保留的，可以通过deployment调度快速回滚
五、小结
● 本文介绍了deployment滚动更新时，deployment、replicaset、pod的细节以及创建过程
● 介绍了deployment版本管理的方式
● 下一小节将会介绍在滚动更新过程中最大可用、liveness以及readiness等
至此，本文结束
在下才疏学浅，有撒汤漏水的，请各位不吝赐教...
经验首页