3-6-2-ReplicaSet和Deployment控制器
控制器主要负责创建 , 管理k8s上的资源 , 如果对应资源不吻合用户定义的资源状态 , 它就会尝试重启或重建的方式让其状态和用户定义的状态吻合 ; 在k8s上控制器的类型有很多 , 比如pod控制 , service控制器 , endpoint控制器等等 ; 不同类型的控制器有着不同的功能和作用 ; 比如pod控制器就是针对pod资源进行管理的控制器 ; 单说pod控制器 , 它也有很多类型 , 根据pod里容器跑的应用程序来分类 , 可以分为有状态应用和无状态应用控制 , 从应用程序是否运行为守护进程我们可以将控制器分为 , 守护进程和非守护进程控制器 ; 其中无状态控制器中最常用的有ReplicaSet控制器和Deployment控制 ; 有状态应用控制器常用的有StatefulSet ; 守护进程控制器最常用的有daemonSet控制器 ; 非守护进程控制器有job控制器 , 对Job类型的控制器 , 如果要周期性执行的有Cronjob控制器 ;
1. ReplicaSet控制器
ReplicaSet控制器的主要作用是确保Pod对象副本数量在任何时刻都能精准满足用户期望的数量 ; 这种控制器启动以后 , 它首先会查找集群中匹配其标签选择器的Pod资源对象 , 当活动pod数量与用户期望的pod数量不吻合时 , 如果多了就删除 , 少了就创建 ; 它创建新pod是靠我们在配置清单中定义的pod模板来创建新pod ;
1.1 创建ReplicaSet控制器
示例 : 定义创建ReplicaSet控制器
root@k8s-master01:/opt/k8s-data/yaml/replicaset# cat replicaset-demo.yaml
apiVersion: apps/v1
kind: ReplicaSet
metadata:
name: replicaset-demo
namespace: default
spec:
replicas: 3
selector:
matchLabels:
app: nginx-pod
template:
metadata:
labels:
app: nginx-pod
spec:
containers:
- name: nginx
image: nginx:1.14-alpine
ports:
- name: http
containerPort: 80
定义ReplicaSet控制器 ,
apiVersion字段的值为apps/v1kind为ReplicaSet, 这两个字段都是固定的 ;metadata中主要定义名称和名称空间 ;spec中主要定义replicas、selector、template;replicas这个字段的值为一个整数 , 表示对应pod的副本数量 ;selector用于定义标签选择器 ; 其值为一个对象 ,matchLabels字段表示精确匹配标签 , 这个字段的值为一个字典 ; 除了精确匹配标签选择器这种方式 , 还有matchExpressions表示使用匹配表达式 , 其值为一个对象 ; 简单说定义标签选择器 ,第一种是
matchLabels, 这种方式就是指定一个或多个标签 , 每个标签就是一个kv键值对 ; 后者matchExpressions是指定一个表达式 , 其值为一个对象 , 这个对象中主要定义key字段 , 这个字段定义key的名称 ;operator定义操作符 ,values定义值 ;key和operator字段的值类型都是字符串 , 其中operator的值有In,NotIn,Exists和DoesNotExist;values是一个字符串列表 ;
其次就是定义
pod模板 , 使用template字段定义 , 该字段的值为一个对象其中metadata字段用于定义模板的元素据信息 , 这个元数据信息必须定义标签属性 ; 通常这个标签属性和选择器中的标签相同 ;spec字段用于定义pod模板的状态 , 最重要的是定义pod里容器的名字 , 镜像等等 ;
应用资源配置清单
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl apply -f replicaset-demo.yaml
replicaset.apps/replicaset-demo created
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get rs
NAME DESIRED CURRENT READY AGE
replicaset-demo 3 3 3 74s
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get rs -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
replicaset-demo 3 3 3 83s nginx nginx:1.14-alpine app=nginx-pod
rs就是ReplicaSet的简写 ; 从上面的信息可以看到对应控制器已经创建 ; 并且当前pod副本数量为3 , 用户期望的数量也为3 , 有3个准备就绪 ;
查看pod
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods -L app
NAME READY STATUS RESTARTS AGE APP
replicaset-demo-fkzrw 1/1 Running 0 25m nginx-pod
replicaset-demo-p9gdx 1/1 Running 0 25m nginx-pod
replicaset-demo-zz7g5 1/1 Running 0 25m nginx-pod可以看到当前default名称空间中创建了3个pod,其标签为nginx-pod ;
测试 ; 更改其中一个pod的标签为ngx,看看对应控制器是否会新建一个标签为nginx-pod的pod
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl label pod/replicaset-demo-fkzrw app=ngx --overwrite
pod/replicaset-demo-fkzrw labeled
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods -L app
NAME READY STATUS RESTARTS AGE APP
replicaset-demo-bhdnf 0/1 ContainerCreating 0 2s nginx-pod
replicaset-demo-fkzrw 1/1 Running 0 28m ngx
replicaset-demo-p9gdx 1/1 Running 0 28m nginx-pod
replicaset-demo-zz7g5 1/1 Running 0 28m nginx-pod
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods -L app
NAME READY STATUS RESTARTS AGE APP
replicaset-demo-bhdnf 1/1 Running 0 5s nginx-pod
replicaset-demo-fkzrw 1/1 Running 0 28m ngx
replicaset-demo-p9gdx 1/1 Running 0 28m nginx-pod
replicaset-demo-zz7g5 1/1 Running 0 28m nginx-pod可以看到当我们把其中一个pod的标签更改为app=ngx后 , 对应控制器又会根据pod模板创建一个新pod ;
1.2 修改标签
测试 : 更改pod标签为app=nginx-pod , 看看对应控制器是否会删除一个pod
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods -L app
NAME READY STATUS RESTARTS AGE APP
replicaset-demo-fkzrw 1/1 Running 0 35m nginx-pod
replicaset-demo-k7h7j 1/1 Running 0 3s nginx-pod
replicaset-demo-p9gdx 1/1 Running 0 35m ngx
replicaset-demo-zz7g5 1/1 Running 0 35m nginx-pod
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl label pod/replicaset-demo-p9gdx app=nginx-pod --overwrite
pod/replicaset-demo-p9gdx labeled
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods -L app
NAME READY STATUS RESTARTS AGE APP
replicaset-demo-fkzrw 1/1 Running 0 35m nginx-pod
replicaset-demo-p9gdx 1/1 Running 0 35m nginx-pod
replicaset-demo-zz7g5 1/1 Running 0 35m nginx-pod
可以看到当集群中有多余用户期望数量的pod标签时 , 对应控制器会把多余的相同标签的pod删除 ; 从上面的测试可以看到ReplicaSet控制器是依靠标签选择器来判断集群中pod的数量是否和用户定义的数量吻合 , 如果不吻合就尝试删除或新建 , 让对应pod数量精确满足用户期望pod数量 ;
查看rs控制器的详细信息
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl describe rs replicaset-demo
Name: replicaset-demo
Namespace: default
Selector: app=nginx-pod
Labels: <none>
Annotations: <none>
Replicas: 3 current / 3 desired
Pods Status: 3 Running / 0 Waiting / 0 Succeeded / 0 Failed
Pod Template:
Labels: app=nginx-pod
Containers:
nginx:
Image: nginx:1.14-alpine
Port: 80/TCP
Host Port: 0/TCP
Environment: <none>
Mounts: <none>
Volumes: <none>
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal SuccessfulCreate 38m replicaset-controller Created pod: replicaset-demo-p9gdx
Normal SuccessfulCreate 38m replicaset-controller Created pod: replicaset-demo-zz7g5
Normal SuccessfulCreate 38m replicaset-controller Created pod: replicaset-demo-fkzrw
Normal SuccessfulCreate 10m replicaset-controller Created pod: replicaset-demo-bhdnf
Normal SuccessfulDelete 4m40s replicaset-controller Deleted pod: replicaset-demo-bhdnf
Normal SuccessfulCreate 4m15s replicaset-controller Created pod: replicaset-demo-ltkl8
Normal SuccessfulDelete 4m1s replicaset-controller Deleted pod: replicaset-demo-ltkl8
Normal SuccessfulCreate 3m24s replicaset-controller Created pod: replicaset-demo-k7h7j
Normal SuccessfulDelete 3m15s replicaset-controller Deleted pod: replicaset-demo-k7h7j1.3 扩展/缩容
扩展/缩减rs控制pod副本数量
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl scale rs replicaset-demo --replicas=6
replicaset.apps/replicaset-demo scaled
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get rs
NAME DESIRED CURRENT READY AGE
replicaset-demo 6 6 6 40m
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl scale rs replicaset-demo --replicas=4
replicaset.apps/replicaset-demo scaled
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get rs
NAME DESIRED CURRENT READY AGE
replicaset-demo 4 4 4 41m
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods
NAME READY STATUS RESTARTS AGE
replicaset-demo-4zzkk 1/1 Running 0 38s
replicaset-demo-fkzrw 1/1 Running 0 41m
replicaset-demo-p9gdx 1/1 Running 0 41m
replicaset-demo-zz7g5 1/1 Running 0 41m
scale也可以对控制器做扩展和缩减pod副本数量 , 除了以上使用命令的方式来变更对应pod副本数量 ; 也可以直接在配置清单中修改replicas字段 , 然后使用apply命令执行配置清单进行修改 ;
修改配置清单中的replicas字段的值来扩展pod副本数量
root@k8s-master01:/opt/k8s-data/yaml/replicaset# cat replicaset-demo.yaml
apiVersion: apps/v1
kind: ReplicaSet
metadata:
name: replicaset-demo
namespace: default
spec:
replicas: 7
selector:
matchLabels:
app: nginx-pod
template:
metadata:
labels:
app: nginx-pod
spec:
containers:
- name: nginx
image: nginx:1.14-alpine
ports:
- name: http
containerPort: 80
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl apply -f replicaset-demo.yaml
replicaset.apps/replicaset-demo configured
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods
NAME READY STATUS RESTARTS AGE
replicaset-demo-272ql 1/1 Running 0 5s
replicaset-demo-4gqb8 1/1 Running 0 5s
replicaset-demo-4zzkk 1/1 Running 0 33m
replicaset-demo-fkzrw 1/1 Running 0 74m
replicaset-demo-nrfbz 1/1 Running 0 5s
replicaset-demo-p9gdx 1/1 Running 0 74m
replicaset-demo-zz7g5 1/1 Running 0 74m
1.4 更新pod版本
1.4.1 更新资源配置清单
方式1 修改资源配置清单中pod模板的版本 , 这里直接将对应yaml文件中的镜像修改为nginx:1.16-alpine, 然后在使用apply命令来执行配置清单
root@k8s-master01:/opt/k8s-data/yaml/replicaset# cat replicaset-demo.yaml
apiVersion: apps/v1
kind: ReplicaSet
metadata:
name: replicaset-demo
namespace: default
spec:
replicas: 7
selector:
matchLabels:
app: nginx-pod
template:
metadata:
labels:
app: nginx-pod
spec:
containers:
- name: nginx
image: nginx:1.16-alpine
ports:
- name: http
containerPort: 80
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl apply -f replicaset-demo.yaml
replicaset.apps/replicaset-demo configured
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get rs -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
replicaset-demo 7 7 7 83m nginx nginx:1.16-alpine app=nginx-pod从上面命令可以看到 , 它显示的镜像版本是1.16的版本 ;
验证 : 查看对应pod , 看看对应pod中容器镜像版本是否变成了1.16
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods
NAME READY STATUS RESTARTS AGE
replicaset-demo-272ql 1/1 Running 0 11m
replicaset-demo-4gqb8 1/1 Running 0 11m
replicaset-demo-4zzkk 1/1 Running 0 45m
replicaset-demo-fkzrw 1/1 Running 0 85m
replicaset-demo-nrfbz 1/1 Running 0 11m
replicaset-demo-p9gdx 1/1 Running 0 85m
replicaset-demo-zz7g5 1/1 Running 0 85m从pod创建的时间来看 , pod没有更新 ;
测试 : 删除一个pod看看对应pod里容器镜像是否会更新
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods
NAME READY STATUS RESTARTS AGE
nginx-demo 1/1 Running 0 10h
nginx-demo2 1/1 Running 0 9h
replicaset-demo-272ql 1/1 Running 0 11m
replicaset-demo-4gqb8 1/1 Running 0 11m
replicaset-demo-4zzkk 1/1 Running 0 45m
replicaset-demo-fkzrw 1/1 Running 0 85m
replicaset-demo-nrfbz 1/1 Running 0 11m
replicaset-demo-p9gdx 1/1 Running 0 85m
replicaset-demo-zz7g5 1/1 Running 0 85m
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl delete pods/replicaset-demo-272ql
pod "replicaset-demo-272ql" deleted
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods
NAME READY STATUS RESTARTS AGE
replicaset-demo-4gqb8 1/1 Running 0 14m
replicaset-demo-4rjx7 1/1 Running 0 34s
replicaset-demo-4zzkk 1/1 Running 0 48m
replicaset-demo-fkzrw 1/1 Running 0 88m
replicaset-demo-nrfbz 1/1 Running 0 14m
replicaset-demo-p9gdx 1/1 Running 0 88m
replicaset-demo-zz7g5 1/1 Running 0 88m
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl describe pods/replicaset-demo-4rjx7 | grep image
Normal Pulling 103s kubelet Pulling image "nginx:1.16-alpine"
Normal Pulled 73s kubelet Successfully pulled image "nginx:1.16-alpine" in 29.986s (29.986s including waiting)可以看到我们删除了一个pod , 对应控制器又新建了一个pod , 对应新建的pod镜像版本就成为了新版本的pod ; 从上面测试情况可以看到 , 对于rs控制器当pod模板中的镜像版本发生更改 , 如果k8s集群上对应pod数量和用户定义的数量吻合 , 此时rs控制器不会更新pod ; 只有新建后的pod才会拥有新版本 ; 也就说如果我们要rs来对pod版本更新 , 就得删除原有老的pod后才会更新 ;
1.4.2 命令更新pod
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl set image rs replicaset-demo nginx=nginx:1.18-alpine
replicaset.apps/replicaset-demo image updated
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get rs -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
replicaset-demo 7 7 7 95m nginx nginx:1.18-alpine app=nginx-pod
root@k8s-master01:/opt/k8s-data/yaml/replicaset# kubectl get pods
NAME READY STATUS RESTARTS AGE
nginx-demo 1/1 Running 0 10h
nginx-demo2 1/1 Running 0 10h
replicaset-demo-4gqb8 1/1 Running 0 21m
replicaset-demo-4rjx7 1/1 Running 0 7m3s
replicaset-demo-4zzkk 1/1 Running 0 54m
replicaset-demo-fkzrw 1/1 Running 0 95m
replicaset-demo-nrfbz 1/1 Running 0 21m
replicaset-demo-p9gdx 1/1 Running 0 95m
replicaset-demo-zz7g5 1/1 Running 0 95m对于rs控制器 , 不管用命令还是修改资源配置清单中pod模板中镜像版本 , 如果有和用户期望数量的pod , 它是不会自动更新pod版本的 ; 只有手动删除老版本pod , 对应新版本pod才会被创建 ;
2. deployment控制器
对于deployment控制来说 , 它的定义方式和rs控制都差不多 , 但deploy控制器的功能要比rs强大 , 它可以实现滚动更新 , 用户手动定义更新策略; 其实deploy控制器是在rs控制器的基础上来管理pod; 也就说我们在创建deploy控制器时 , 它自动会创建一个rs控制器; 其中使用deployment控制器创建的pod名称是由deploy控制器名称加上“-”pod模板hash名称加上“-”随机字符串; 而对应rs控制器的名称恰好就是deploy控制器名称加“-”pod模板hash; 即pod名称就为rs控制器名称加“-”随机字符串;
2.1 创建deployment控制器
示例：创建deployment控制器
root@k8s-master01:/opt/k8s-data/yaml/deployment# cat deployment-demo.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: deploy-demo
namespace: default
spec:
replicas: 3
selector:
matchLabels:
app: ngx-dep-pod
template:
metadata:
labels:
app: ngx-dep-pod
spec:
containers:
- name: nginx
image: nginx:1.14-alpine
ports:
- name: http
containerPort: 80应用配置清单
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl apply -f deployment-demo.yaml
deployment.apps/deploy-demo created
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get deploy -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
deploy-demo 3/3 3 3 8s nginx nginx:1.14-alpine app=ngx-dep-pod验证 : 查看是否有rs控制器创建
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get rs
NAME DESIRED CURRENT READY AGE
deploy-demo-57f99896b5 3 3 3 54s可以看到有一个deploy-demo-57f99896b5控制器被创建 ;
验证：查看pod , 看看对应pod名称是否有rs控制器名称加“-”一串随机字符串
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get pods
NAME READY STATUS RESTARTS AGE
deploy-demo-57f99896b5-g89gg 1/1 Running 0 2m34s
deploy-demo-57f99896b5-mktrc 1/1 Running 0 2m34s
deploy-demo-57f99896b5-pnqz4 1/1 Running 0 2m34s可以看到有3个pod的名称是deploy-demo-57f99896b5-加随机字符串 ;
2.2 更新pod版本
2.2.1 更新配置清单
这里通过编辑yaml文件, 调整image镜像版本。这里直接将image版本从1.14-alpine修改为1.16-alpine
root@k8s-master01:/opt/k8s-data/yaml/deployment# cat deployment-demo.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: deploy-demo
namespace: default
spec:
replicas: 3
selector:
matchLabels:
app: ngx-dep-pod
template:
metadata:
labels:
app: ngx-dep-pod
spec:
containers:
- name: nginx
image: nginx:1.16-alpine
ports:
- name: http
containerPort: 80
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl apply -f deployment-demo.yaml
deployment.apps/deploy-demo configured
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get deployment -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
deploy-demo 3/3 2 3 9m16s nginx nginx:1.16-alpine app=ngx-dep-pod
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get pod
NAME READY STATUS RESTARTS AGE
deploy-demo-55c5b97f5-52lk6 1/1 Running 0 8s
deploy-demo-55c5b97f5-6fx2z 1/1 Running 0 34s
deploy-demo-55c5b97f5-pzml4 1/1 Running 0 60s可以看到deploy控制器只要更改了pod模板中镜像版本 , 对应pod会自动更新 ;
2.2.2 命令更新pod
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl set image deployment deploy-demo nginx=nginx:1.18-alpine
deployment.apps/deploy-demo image updated
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get deployment -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
deploy-demo 3/3 1 3 15m nginx nginx:1.18-alpine app=ngx-dep-pod
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get pod
NAME READY STATUS RESTARTS AGE
deploy-demo-777c6bc997-czhkr 1/1 Running 0 66s
deploy-demo-777c6bc997-rfv4s 1/1 Running 0 88s
deploy-demo-777c6bc997-s79nr 1/1 Running 0 39s可以看到deploy控制器 , 只要修改了pod模板中镜像的版本 , 对应pod就会随之滚动更新到我们指定的版本;
2.3更新历史版本
2.3.1 查看rs历史版本
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get rs -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
deploy-demo-55c5b97f5 0 0 0 11m nginx nginx:1.16-alpine app=ngx-dep-pod,pod-template-hash=55c5b97f5
deploy-demo-57f99896b5 0 0 0 19m nginx nginx:1.14-alpine app=ngx-dep-pod,pod-template-hash=57f99896b5
deploy-demo-777c6bc997 3 3 3 4m18s nginx nginx:1.18-alpine app=ngx-dep-pod,pod-template-hash=777c6bc997deploy控制器的更新pod版本操作 , 它会记录rs的所有历史版本; 因为只要pod模板的hash值发生变化 , 对应的rs就会重新被创建一遍 , 不同于rs控制器 , 历史版本的rs上没有pod运行 , 只有当前版本的rs上才会运行pod ;
2.3.2 查看更新历史记录
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout history deploy/deploy-demo
deployment.apps/deploy-demo
REVISION CHANGE-CAUSE
1 <none>
2 <none>
3 <none>这里可以看到有3个版本 , 没有记录对应的原因; 这是因为我们在更新pod版本是没有记录; 要想记录器更新原因 , 可以在对应名后面加--record选项即可;
示例 : 记录更新操作命令到更新历史记录
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record
Flag --record has been deprecated, --record will be removed in the future
deployment.apps/deploy-demo image updated
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout history deploy/deploy-demo
deployment.apps/deploy-demo
REVISION CHANGE-CAUSE
2 <none>
3 <none>
4 kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record=true可以看到更新操作时加上--record选项后 , 再次查看更新历史记录 , 就能显示对应的更新命令;
新版本已经弃用了该参数, 新版本的annotation会沿用上一个版本的内容。
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl set image deployment deploy-demo nginx=nginx:1.16-alpine
deployment.apps/deploy-demo image updated
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout history deploy/deploy-demo
deployment.apps/deploy-demo
REVISION CHANGE-CAUSE
3 <none>
4 kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record=true
5 kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record=true取而代之的是命令如下 , 这个命令需要每次更新deployment后都执行一次
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl annotate deployment/deploy-demo kubernetes.io/change-cause="image updated to 1.16-alpine"
deployment.apps/deploy-demo annotated
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout history deploy/deploy-demo
deployment.apps/deploy-demo
REVISION CHANGE-CAUSE
3 <none>
4 kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record=true
5 image updated to 1.16-alpine可以发现更新原因已经发生变化
2.3.3回到上一个版本
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get deploy -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
deploy-demo 3/3 3 3 52m nginx nginx:1.16-alpine app=ngx-dep-pod
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get rs -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
deploy-demo-55c5b97f5 3 3 3 44m nginx nginx:1.16-alpine app=ngx-dep-pod,pod-template-hash=55c5b97f5
deploy-demo-57f99896b5 0 0 0 52m nginx nginx:1.14-alpine app=ngx-dep-pod,pod-template-hash=57f99896b5
deploy-demo-777c6bc997 0 0 0 37m nginx nginx:1.18-alpine app=ngx-dep-pod,pod-template-hash=777c6bc997
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout history deploy/deploy-demo
deployment.apps/deploy-demo
REVISION CHANGE-CAUSE
4 kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record=true
6 image updated to 1.16-alpine
7 image updated to 1.16-alpine
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout undo deploy/deploy-demo
deployment.apps/deploy-demo rolled back
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout history deploy/deploy-demo
deployment.apps/deploy-demo
REVISION CHANGE-CAUSE
4 kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record=true
7 image updated to 1.16-alpine
8 image updated to 1.16-alpine
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get deploy -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
deploy-demo 3/3 3 3 56m nginx nginx:1.18-alpine app=ngx-dep-pod
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get rs -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
deploy-demo-55c5b97f5 0 0 0 48m nginx nginx:1.16-alpine app=ngx-dep-pod,pod-template-hash=55c5b97f5
deploy-demo-57f99896b5 0 0 0 56m nginx nginx:1.14-alpine app=ngx-dep-pod,pod-template-hash=57f99896b5
deploy-demo-777c6bc997 3 3 3 41m nginx nginx:1.18-alpine app=ngx-dep-pod,pod-template-hash=777c6bc997可以看到执行了kubectl rollout undo deploy/deploy-demo命令后 , 对应版本从1.16就回滚到1.18的版本了; 对应更新历史记录也把1.18版本更新为当前最新记录;
2.3.4 回滚到指定历史记录版本
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout history deploy/deploy-demo
deployment.apps/deploy-demo
REVISION CHANGE-CAUSE
4 kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record=true
7 image updated to 1.16-alpine
8 image updated to 1.16-alpine
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout undo deploy/deploy-demo --to-revision=4
deployment.apps/deploy-demo rolled back
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout history deploy/deploy-demo
deployment.apps/deploy-demo
REVISION CHANGE-CAUSE
7 image updated to 1.16-alpine
8 image updated to 1.16-alpine
9 kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record=true
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get deploy -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
deploy-demo 3/3 3 3 61m nginx nginx:1.14-alpine app=ngx-dep-pod
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get rs -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
deploy-demo-55c5b97f5 0 0 0 53m nginx nginx:1.16-alpine app=ngx-dep-pod,pod-template-hash=55c5b97f5
deploy-demo-57f99896b5 3 3 3 62m nginx nginx:1.14-alpine app=ngx-dep-pod,pod-template-hash=57f99896b5
deploy-demo-777c6bc997 0 0 0 46m nginx nginx:1.18-alpine app=ngx-dep-pod,pod-template-hash=777c6bc997指定要回滚到某个历史记录的版本 , 可以使用--to-revision选项来指定历史记录的编号 ;
查看deploy控制器的详细信息
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl describe deploy deploy-demo
Name: deploy-demo
Namespace: default
CreationTimestamp: Tue, 04 Jun 2024 17:03:17 +0000
Labels: <none>
Annotations: deployment.kubernetes.io/revision: 9
kubernetes.io/change-cause: kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record=true
Selector: app=ngx-dep-pod
Replicas: 3 desired | 3 updated | 3 total | 3 available | 0 unavailable
StrategyType: RollingUpdate
MinReadySeconds: 0
RollingUpdateStrategy: 25% max unavailable, 25% max surge
Pod Template:
Labels: app=ngx-dep-pod
Containers:
nginx:
Image: nginx:1.14-alpine
Port: 80/TCP
Host Port: 0/TCP
Environment: <none>
Mounts: <none>
Volumes: <none>
Conditions:
Type Status Reason
---- ------ ------
Available True MinimumReplicasAvailable
Progressing True NewReplicaSetAvailable
OldReplicaSets: deploy-demo-55c5b97f5 (0/0 replicas created), deploy-demo-777c6bc997 (0/0 replicas created)
NewReplicaSet: deploy-demo-57f99896b5 (3/3 replicas created)
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal ScalingReplicaSet 55m deployment-controller Scaled up replica set deploy-demo-55c5b97f5 to 1
Normal ScalingReplicaSet 55m deployment-controller Scaled down replica set deploy-demo-57f99896b5 to 2 from 3
Normal ScalingReplicaSet 55m deployment-controller Scaled up replica set deploy-demo-55c5b97f5 to 2 from 1
Normal ScalingReplicaSet 54m deployment-controller Scaled down replica set deploy-demo-57f99896b5 to 1 from 2
Normal ScalingReplicaSet 54m deployment-controller Scaled down replica set deploy-demo-57f99896b5 to 0 from 1
Normal ScalingReplicaSet 48m deployment-controller Scaled up replica set deploy-demo-777c6bc997 to 1
Normal ScalingReplicaSet 31m (x2 over 54m) deployment-controller Scaled up replica set deploy-demo-55c5b97f5 to 3 from 2
Normal ScalingReplicaSet 22m deployment-controller Scaled up replica set deploy-demo-777c6bc997 to 3 from 2
Normal ScalingReplicaSet 20m deployment-controller Scaled up replica set deploy-demo-55c5b97f5 to 1 from 0
Normal ScalingReplicaSet 20m (x21 over 48m) deployment-controller (combined from similar events): Scaled down replica set deploy-demo-777c6bc997 to 2 from 3
Normal ScalingReplicaSet 9m45s deployment-controller Scaled up replica set deploy-demo-777c6bc997 to 1 from 0
Normal ScalingReplicaSet 9m43s (x2 over 48m) deployment-controller Scaled down replica set deploy-demo-55c5b97f5 to 2 from 3
Normal ScalingReplicaSet 3m55s deployment-controller Scaled up replica set deploy-demo-57f99896b5 to 1 from 0查看deploy控制器的详细信息 , 可以看到对应pod模板 , 回滚的过程 , 以及默认更新策略等等信息 ;
2.3.5 自定义滚动更新策略
root@k8s-master01:/opt/k8s-data/yaml/deployment# cat deployment-demo.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: deploy-demo
namespace: default
spec:
revisionHistoryLimit: 10
replicas: 3
selector:
matchLabels:
app: ngx-dep-pod
template:
metadata:
labels:
app: ngx-dep-pod
spec:
containers:
- name: nginx
image: nginx:1.16-alpine
ports:
- name: http
containerPort: 80
strategy:
type: RollingUpdate
rollingUpdate:
maxSurge: 2
maxUnavailable: 1
minReadySeconds: 5 定义滚动更新策略需要使用strategy这个字段 , 这个字段的值是一个对象 , 其中type是指定更新策略 , 其策略有两种 ,
第一种是
Recreate, 这种策略更新方式是新建一个新版pod, 然后再删除一个旧版pod以这种方式滚动更新 ;第二种是
RollingUpdate, 这种策略是用于我们手动指定的策略 ;maxSurge表示最大允许超出用户期望的pod数量（即更新时允许新建超出用户期望的pod数量）maxUnavailable表示最大允许少于用于期望的pod数量（即更新时可以一次删除几个旧版pod）
最后minReadySeconds字段不是定义更新策略的 , 它是spec中的一个字段 , 用于限定pod最小就绪时长 ; 以上更新策略表示 , 使用RollingUpdate类型策略 , 并指定最大新建pod超出用户期望pod数量为2个 , 最大允许少于用户期望pod数量为1个 ; pod最小就绪时间为5秒 ;
应用配置清单
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl apply -f deployment-demo.yaml
deployment.apps/deploy-demo configured
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl describe deploy/deploy-demo
Name: deploy-demo
Namespace: default
CreationTimestamp: Tue, 04 Jun 2024 17:03:17 +0000
Labels: <none>
Annotations: deployment.kubernetes.io/revision: 10
kubernetes.io/change-cause: kubectl set image deployment deploy-demo nginx=nginx:1.14-alpine --record=true
Selector: app=ngx-dep-pod
Replicas: 3 desired | 3 updated | 3 total | 3 available | 0 unavailable
StrategyType: RollingUpdate
MinReadySeconds: 5
RollingUpdateStrategy: 1 max unavailable, 2 max surge
Pod Template:
Labels: app=ngx-dep-pod
Containers:
nginx:
Image: nginx:1.16-alpine
Port: 80/TCP
Host Port: 0/TCP
Environment: <none>
Mounts: <none>
Volumes: <none>
Conditions:
Type Status Reason
---- ------ ------
Available True MinimumReplicasAvailable
Progressing True NewReplicaSetAvailable
OldReplicaSets: deploy-demo-57f99896b5 (0/0 replicas created), deploy-demo-777c6bc997 (0/0 replicas created)
NewReplicaSet: deploy-demo-55c5b97f5 (3/3 replicas created)
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal ScalingReplicaSet 58m deployment-controller Scaled up replica set deploy-demo-777c6bc997 to 1
Normal ScalingReplicaSet 40m (x2 over 64m) deployment-controller Scaled up replica set deploy-demo-55c5b97f5 to 3 from 2
Normal ScalingReplicaSet 31m deployment-controller Scaled up replica set deploy-demo-777c6bc997 to 3 from 2
Normal ScalingReplicaSet 30m deployment-controller Scaled up replica set deploy-demo-55c5b97f5 to 1 from 0
Normal ScalingReplicaSet 30m (x21 over 57m) deployment-controller (combined from similar events): Scaled down replica set deploy-demo-777c6bc997 to 2 from 3
Normal ScalingReplicaSet 19m deployment-controller Scaled up replica set deploy-demo-777c6bc997 to 1 from 0
Normal ScalingReplicaSet 19m (x2 over 57m) deployment-controller Scaled down replica set deploy-demo-55c5b97f5 to 2 from 3
Normal ScalingReplicaSet 13m deployment-controller Scaled up replica set deploy-demo-57f99896b5 to 1 from 0
Normal ScalingReplicaSet 26s (x2 over 64m) deployment-controller Scaled down replica set deploy-demo-57f99896b5 to 2 from 3
Normal ScalingReplicaSet 26s deployment-controller Scaled up replica set deploy-demo-55c5b97f5 to 2 from 0可以看到对应deploy控制器的更新策略已经更改为我们定义的策略 ; 为了能够看出更新的效果 , 我们这里先手动把pod数量调整为10个 ;
2.3.6. 扩展pod副本数量
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl scale deploy/deploy-demo --replicas=10
deployment.apps/deploy-demo scaled
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get pods
NAME READY STATUS RESTARTS AGE
deploy-demo-55c5b97f5-7b6fx 1/1 Running 0 10s
deploy-demo-55c5b97f5-8l5lr 1/1 Running 0 10s
deploy-demo-55c5b97f5-9lkj6 1/1 Running 0 10s
deploy-demo-55c5b97f5-dwj2k 1/1 Running 0 9m22s
deploy-demo-55c5b97f5-fqtm8 1/1 Running 0 10s
deploy-demo-55c5b97f5-gfvnv 1/1 Running 0 9m22s
deploy-demo-55c5b97f5-j86kr 1/1 Running 0 10s
deploy-demo-55c5b97f5-nkq89 1/1 Running 0 10s
deploy-demo-55c5b97f5-wn5ng 1/1 Running 0 9m22s
deploy-demo-55c5b97f5-zfdb4 1/1 Running 0 10s查看更新过程
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get pods -w
NAME READY STATUS RESTARTS AGE
deploy-demo-55c5b97f5-gfvnv 1/1 Running 0 11m
deploy-demo-55c5b97f5-wn5ng 1/1 Running 0 11m
deploy-demo-777c6bc997-9xfhs 1/1 Running 0 3s
deploy-demo-777c6bc997-hk6ff 1/1 Running 0 3s
deploy-demo-777c6bc997-k95xj 1/1 Running 0 3s
nginx-demo 1/1 Running 0 12h
nginx-demo2 1/1 Running 0 11h
deploy-demo-55c5b97f5-gfvnv 1/1 Terminating 0 11m
deploy-demo-55c5b97f5-wn5ng 1/1 Terminating 0 11m
deploy-demo-55c5b97f5-wn5ng 0/1 Terminating 0 11m
deploy-demo-55c5b97f5-gfvnv 0/1 Terminating 0 11m
deploy-demo-55c5b97f5-wn5ng 0/1 Terminating 0 11m
deploy-demo-55c5b97f5-wn5ng 0/1 Terminating 0 11m
deploy-demo-55c5b97f5-wn5ng 0/1 Terminating 0 11m
.....使用-w选项可以一直跟踪查看pod变化过程 ; 从上面的监控信息可以看到 , 在更新时 , 首先是将三个pod标记为pending状态 , 然后先删除一个pod , 然后再创建两个pod ; 然后又创建一个 , 再删除3个 , 一次进行 ; 不管怎么删除和新建 , 对应新旧pod的数量最少要有9个 , 最大不超过12个 ;
使用暂停更新实现金丝雀发布
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl set image deploy/deploy-demo nginx=nginx:1.16-alpine && kubectl rollout pause deploy/deploy-demo
deployment.apps/deploy-demo image updated
deployment.apps/deploy-demo paused以上命令会根据我们定义的更新策略 , 先删除一个pod , 然后再创建3个新版pod , 然后更新操作就暂停了 ; 此时对应pod只更新了1个 , 然后新建了2个新pod , 总共就有12个pod ;
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
deploy-demo-55c5b97f5-4v4jm 1/1 Running 0 3s 172.20.3.111 k8s-node01.cbear.top <none> <none>
deploy-demo-55c5b97f5-h2gsr 1/1 Running 0 3s 172.20.4.86 k8s-node02.cbear.top <none> <none>
deploy-demo-55c5b97f5-wg52h 1/1 Running 0 3s 172.20.5.72 k8s-node03.cbear.top <none> <none>
deploy-demo-57f99896b5-24vpg 1/1 Running 0 70s 172.20.4.84 k8s-node02.cbear.top <none> <none>
deploy-demo-57f99896b5-2j847 1/1 Running 0 70s 172.20.3.108 k8s-node01.cbear.top <none> <none>
deploy-demo-57f99896b5-98hcj 1/1 Running 0 63s 172.20.4.85 k8s-node02.cbear.top <none> <none>
deploy-demo-57f99896b5-9dwx2 1/1 Running 0 76s 172.20.5.69 k8s-node03.cbear.top <none> <none>
deploy-demo-57f99896b5-cxmp6 1/1 Running 0 76s 172.20.4.83 k8s-node02.cbear.top <none> <none>
deploy-demo-57f99896b5-lp2hh 1/1 Running 0 63s 172.20.5.71 k8s-node03.cbear.top <none> <none>
deploy-demo-57f99896b5-qcb7w 1/1 Running 0 76s 172.20.3.107 k8s-node01.cbear.top <none> <none>
deploy-demo-57f99896b5-tkngt 1/1 Running 0 63s 172.20.3.109 k8s-node01.cbear.top <none> <none>
deploy-demo-57f99896b5-w8hhz 1/1 Running 0 70s 172.20.5.70 k8s-node03.cbear.top <none> <none>之所以多两个是因为我们在更新策略中定义允许最大超出用户期望2个pod ;
恢复更新
root@k8s-master01:/opt/k8s-data/yaml/deployment# kubectl rollout resume deploy/deploy-demo && kubectl rollout status deploy/deploy-demo
deployment.apps/deploy-demo resumed
Waiting for deployment "deploy-demo" rollout to finish: 3 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 3 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 6 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 6 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 6 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 6 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 6 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 6 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 9 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 9 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 9 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 9 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 9 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 9 out of 10 new replicas have been updated...
Waiting for deployment "deploy-demo" rollout to finish: 9 of 10 updated replicas are available...
Waiting for deployment "deploy-demo" rollout to finish: 9 of 10 updated replicas are available...
deployment "deploy-demo" successfully rolled outresume表示恢复刚才暂停的更新操作 ; status是用来查看对应更新过程 ;
如果“金丝雀”遇险，这时候需要恢复滚动更新并立即回滚：
$ kubectl rollout resume deoloyments my-app-deployment \
&& kubectl rollout undo deoloyments my-app-deployment回滚完成后验证 Deployment 管理的 ReplicaSet 控制器对象是否已恢复到指定的历史版本以确保回滚正常。在 kubectl rollout undo 后跟上 --to-revision=N 选项指定历史版本序号即可回滚至历史特定版本。
参考链接
容器编排系统k8s之ReplicaSet和Deployment控制器 - Linux-1874 - 博客园 (cnblogs.com)