deployment
deployment基础
上一节试验了replicaset控制器，它可以保证设置的pod数量，但是在升级时，却做不到自动升级，kubectl apply -f xxxx，并不能实时生效。
今天看看deployment，一种比replicaset更高级的声明式资源。它具有滚动更新、平滑升级、蓝绿发布、金丝雀发布等优点。可以直接通过kubectl apply -f xxx完成所有操作。
Deployment控制器是建立在rs之上的一个控制器，定义一个deployment控制器就会增加一个replicaset，它可以管理多个rs，每次更新镜像版本，都会生成一个新的rs，把旧的rs替换掉，多个rs同时存在，但是只有一个rs运行。
deployment yaml文件语法
apiVersion: apps/v1
kind: Deployment
metadata:
name: first-deployment
labels:
func: test-deployment
spec:
replicas: 3 # deployment管理的POD数量
selector: # deployment管理哪些POD
matchLabels:
app: nginx
template: # pod的模板信息
metadata:
name: nginx-test
labels:
app: nginx
spec:
tolerations: # 配置容忍度
- effect: "NoSchedule"
key: "can-run-pods"
operator: "Equal"
value: "master-node"
- effect: "NoSchedule"
key: "node-role.kubernetes.io/control-plane"
operator: "Equal"
value:
containers:
- name: nginx-test
image: nginx:stable-alpine-perl # container所用镜像
imagePullPolicy: IfNotPresent # container镜像拉取策略
env:
- name: VERSION # 定义环境变量
value: 1.1.0
lifecycle: # POD声明周期，启动后钩子
postStart:
exec:
command: ["/bin/sh","-c","echo the version is $VERSION, this is a nginx server in container, and the ip is $(ip add show eth0 | grep -w inet | awk '{print $2}') > /usr/share/nginx/html/index.html" ] # 启动后钩子，生成定制的主页文件
startupProbe: # 启动探测
periodSeconds: 5
initialDelaySeconds: 20
timeoutSeconds: 10
httpGet:
scheme: HTTP
port: 80
path: /
livenessProbe: # 存活探测
periodSeconds: 5
initialDelaySeconds: 20
timeoutSeconds: 10
httpGet:
scheme: HTTP
port: 80
path: /
readinessProbe: # 就绪探测
periodSeconds: 5
initialDelaySeconds: 20
timeoutSeconds: 10
httpGet:
scheme: HTTP
port: 80
path: /
生成了与replicas数量相等的POD
[root@master-worker-node-1 deployment]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
first-deployment-66494996ff-58x5z 1/1 Running 0 5m48s 10.244.54.58 only-worker-node-4 <none> <none>
first-deployment-66494996ff-clkwk 1/1 Running 0 5m48s 10.244.132.211 master-worker-node-2 <none> <none>
first-deployment-66494996ff-t7lc8 1/1 Running 0 5m48s 10.244.31.48 only-worker-node-3 <none> <none>
同时会创建一个replicaset资源
[root@master-worker-node-1 deployment]# kubectl get replicaset -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
first-deployment-66494996ff 3 3 3 7m16s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
[root@master-worker-node-1 deployment]# kubectl get deployment -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
first-deployment 3/3 3 3 8m53s nginx-test nginx:stable-alpine-perl app=nginx
pod的名字是以deployment名字+随机的RS编号+随机的POD编号组成
deployment的扩容和缩容
通过kubectl edit 修改或者通过修改deploy yaml文件后通过apply声明。
将deployment扩容为5
[root@master-worker-node-1 deployment]# diff deploy.yaml deploy-expansion.yaml
8c8
< replicas: 3
---
> replicas: 5
[root@master-worker-node-1 deployment]# kubectl apply -f deploy-expansion.yaml
deployment.apps/first-deployment configured
[root@master-worker-node-1 deployment]# kubectl get deployment -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
first-deployment 5/5 5 5 22m nginx-test nginx:stable-alpine-perl app=nginx
[root@master-worker-node-1 deployment]#
[root@master-worker-node-1 deployment]# kubectl get replicaset -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
first-deployment-66494996ff 5 5 5 22m nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
[root@master-worker-node-1 deployment]#
[root@master-worker-node-1 deployment]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
first-deployment-66494996ff-58x5z 1/1 Running 0 22m 10.244.54.58 only-worker-node-4 <none> <none>
first-deployment-66494996ff-clkwk 1/1 Running 0 22m 10.244.132.211 master-worker-node-2 <none> <none>
first-deployment-66494996ff-j78rp 1/1 Running 0 3m35s 10.244.31.49 only-worker-node-3 <none> <none>
first-deployment-66494996ff-t7lc8 1/1 Running 0 22m 10.244.31.48 only-worker-node-3 <none> <none>
first-deployment-66494996ff-w2fpc 1/1 Running 0 45s 10.244.54.59 only-worker-node-4 <none> <none>
将deployment缩容为2
[root@master-worker-node-1 deployment]# kubectl edit deployment first-deployment
deployment.apps/first-deployment edited
[root@master-worker-node-1 deployment]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
first-deployment-66494996ff-clkwk 1/1 Running 0 24m 10.244.132.211 master-worker-node-2 <none> <none>
first-deployment-66494996ff-t7lc8 1/1 Running 0 24m 10.244.31.48 only-worker-node-3 <none> <none>
[root@master-worker-node-1 deployment]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
first-deployment-66494996ff-bnfcq 1/1 Running 0 39s 10.244.54.7 only-worker-node-4 <none> <none>
first-deployment-66494996ff-f5257 1/1 Running 0 39s 10.244.132.221 master-worker-node-2 <none> <none>
first-deployment-66494996ff-vj49c 1/1 Running 0 39s 10.244.31.59 only-worker-node-3 <none> <none>
[root@master-worker-node-1 deployment]# kubectl scale deployment first-deployment --replicas 5
deployment.apps/first-deployment scaled
[root@master-worker-node-1 deployment]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
first-deployment-66494996ff-bnfcq 1/1 Running 0 115s 10.244.54.7 only-worker-node-4 <none> <none>
first-deployment-66494996ff-f5257 1/1 Running 0 115s 10.244.132.221 master-worker-node-2 <none> <none>
first-deployment-66494996ff-p44wx 1/1 Running 0 63s 10.244.31.61 only-worker-node-3 <none> <none>
first-deployment-66494996ff-spgc8 1/1 Running 0 63s 10.244.54.8 only-worker-node-4 <none> <none>
first-deployment-66494996ff-vj49c 1/1 Running 0 115s 10.244.31.59 only-worker-node-3 <none> <none>
[root@master-worker-node-1 deployment]#
使用的deployment实现滚动更新
deployment滚动升级涉及到升级策略：
kubectl explain deployment.spec.strategy
rollingUpdate：
type: Possible enum values:
- `"Recreate"` Kill all existing pods before creating new ones. 删除所有，再重建所有。
- `"RollingUpdate"` Replace the old ReplicaSets by new one using rolling update i.e gradually scale down the old ReplicaSets and scale up the new one. 删除一个，随之重建一个。
kubectl explain deployment.spec.strategy.rollingUpdate
maxSurge：The maximum number of pods that can be scheduled above the desired number of pods. 默认25%
maxUnavailable： The maximum number of pods that can be unavailable during the update. 默认25%
deployment升级过程中的POD数量取值范围：期望数量-maxUnavailable≤pod数量≤期望数量+maxSurge
不配置升级策略进行滚动升级
# 当前环境
[root@master-worker-node-1 deployment]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
first-deployment-66494996ff-clkwk 1/1 Running 0 52m 10.244.132.211 master-worker-node-2 <none> <none>
first-deployment-66494996ff-t7lc8 1/1 Running 0 52m 10.244.31.48 only-worker-node-3 <none> <none>
[root@master-worker-node-1 deployment]# kubectl get pods -o wide | awk '{print $6}'
IP
10.244.132.211
10.244.31.48
[root@master-worker-node-1 deployment]# kubectl get pods -o wide | awk '{print $6}' | grep -v IP
10.244.132.211
10.244.31.48
[root@master-worker-node-1 deployment]# kubectl get pods -o wide | awk '{print $6}' | grep -v IP | xargs curl
the version is 1.1.0, this is a nginx server in container, and the ip is 10.244.132.211/32
the version is 1.1.0, this is a nginx server in container, and the ip is 10.244.31.48/32
# 修改一个版本号，模拟镜像变化
[root@master-worker-node-1 deployment]# diff deploy.yaml deploy-update.yaml
33c33
< value: 1.1.0
---
> value: 1.2.0
[root@master-worker-node-1 deployment]# kubectl apply -f deploy-update.yaml
deployment.apps/first-deployment configured
[root@master-worker-node-1 deployment]# kubectl get deployment -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
first-deployment 3/3 2 3 3m23s nginx-test nginx:stable-alpine-perl app=nginx
[root@master-worker-node-1 deployment]# kubectl get replicaset -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
first-deployment-66494996ff 0 0 0 3m54s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
first-deployment-7d6b98897d 3 3 3 83s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=7d6b98897d
[root@master-worker-node-1 deployment]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
first-deployment-7d6b98897d-t8vdf 1/1 Running 0 66s 10.244.54.6 only-worker-node-4 <none> <none>
first-deployment-7d6b98897d-w6lvc 1/1 Running 0 93s 10.244.31.57 only-worker-node-3 <none> <none>
first-deployment-7d6b98897d-wrxxc 1/1 Running 0 40s 10.244.132.219 master-worker-node-2 <none> <none>
# 主页已经更新到了version 1.2.0
[root@master-worker-node-1 deployment]# kubectl get pods -o wide | awk '{print $6}' | grep -v IP | xargs curl
the version is 1.2.0, this is a nginx server in container, and the ip is 10.244.54.6/32
the version is 1.2.0, this is a nginx server in container, and the ip is 10.244.31.57/32
the version is 1.2.0, this is a nginx server in container, and the ip is 10.244.132.219/32
完成上面的更新以后，发现pod对应的replicaset也发生的相应的变化。deployment进行了更新，同步的，也会产生一个对应的rs。既然如此就可以通过之前rs回复到之前的情况。
查看deployment的历史版本
[root@master-worker-node-1 ~]# kubectl rollout history deployment first-deployment
deployment.apps/first-deployment
REVISION CHANGE-CAUSE
1 <none>
2 <none>
回滚到对应的版本
[root@master-worker-node-1 ~]# kubectl rollout undo deployment first-deployment --to-revision=1
deployment.apps/first-deployment rolled back
升级过程中pod的变化情况
[root@master-worker-node-1 deployment]# kubectl get pods -o wide -w
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
first-deployment-7d6b98897d-t8vdf 1/1 Running 0 5m28s 10.244.54.6 only-worker-node-4 <none> <none>
first-deployment-7d6b98897d-w6lvc 1/1 Running 0 5m55s 10.244.31.57 only-worker-node-3 <none> <none>
first-deployment-7d6b98897d-wrxxc 1/1 Running 0 5m2s 10.244.132.219 master-worker-node-2 <none> <none>
first-deployment-66494996ff-47xhv 0/1 Pending 0 0s <none> <none> <none> <none>
first-deployment-66494996ff-47xhv 0/1 Pending 0 0s <none> only-worker-node-3 <none> <none>
first-deployment-66494996ff-47xhv 0/1 ContainerCreating 0 1s <none> only-worker-node-3 <none> <none>
first-deployment-66494996ff-47xhv 0/1 ContainerCreating 0 2s <none> only-worker-node-3 <none> <none>
first-deployment-66494996ff-47xhv 0/1 Running 0 3s 10.244.31.58 only-worker-node-3 <none> <none>
first-deployment-66494996ff-47xhv 0/1 Running 0 26s 10.244.31.58 only-worker-node-3 <none> <none>
first-deployment-66494996ff-47xhv 1/1 Running 0 26s 10.244.31.58 only-worker-node-3 <none> <none>
first-deployment-7d6b98897d-w6lvc 1/1 Terminating 0 8m28s 10.244.31.57 only-worker-node-3 <none> <none>
first-deployment-66494996ff-qhbfc 0/1 Pending 0 1s <none> <none> <none> <none>
first-deployment-66494996ff-qhbfc 0/1 Pending 0 1s <none> only-worker-node-4 <none> <none>
first-deployment-66494996ff-qhbfc 0/1 ContainerCreating 0 1s <none> only-worker-node-4 <none> <none>
first-deployment-7d6b98897d-w6lvc 1/1 Terminating 0 8m29s 10.244.31.57 only-worker-node-3 <none> <none>
first-deployment-66494996ff-qhbfc 0/1 ContainerCreating 0 3s <none> only-worker-node-4 <none> <none>
first-deployment-7d6b98897d-w6lvc 0/1 Terminating 0 8m31s 10.244.31.57 only-worker-node-3 <none> <none>
first-deployment-7d6b98897d-w6lvc 0/1 Terminating 0 8m31s 10.244.31.57 only-worker-node-3 <none> <none>
first-deployment-7d6b98897d-w6lvc 0/1 Terminating 0 8m31s 10.244.31.57 only-worker-node-3 <none> <none>
first-deployment-66494996ff-qhbfc 0/1 Running 0 4s 10.244.54.4 only-worker-node-4 <none> <none>
设置升级策略进行滚动升级
# 默认的升级策略，滚动式，最大最小均为25%
[root@master-worker-node-1 deployment]# kubectl get deployment first-deployment -o yaml | grep -A 5 strategy
strategy:
rollingUpdate:
maxSurge: 25%
maxUnavailable: 25%
type: RollingUpdate
[root@master-worker-node-1 deployment]# kubectl apply -f deploy-update-self.yaml
deployment.apps/my-deployment created
[root@master-worker-node-1 deployment]# kubectl get deployment -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
my-deployment 3/3 3 3 29s nginx-test nginx:stable-alpine-perl app=nginx
[root@master-worker-node-1 deployment]# kubectl get deployment my-deployment -o yaml | grep -A 5 strategy: -w
strategy:
rollingUpdate:
maxSurge: 2
maxUnavailable: 1
type: RollingUpdate
template:
[root@master-worker-node-1 deployment]#
#在线更新deployment
[root@master-worker-node-1 deployment]# kubectl apply -f deploy-update-self-2.yaml
deployment.apps/my-deployment configured
[root@master-worker-node-1 deployment]# kubectl get deployment my-deployment -o wide
NAME READY UP-TO-DATE AVAILABLE AGE CONTAINERS IMAGES SELECTOR
my-deployment 3/3 3 3 4m10s nginx-test nginx:stable-alpine-perl app=nginx
# 观察rs的状态，更好的观察pod变化
[root@master-worker-node-1 ~]# kubectl get rs -o wide -w
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
my-deployment-66494996ff 3 3 3 3m9s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
my-deployment-7d6b98897d 2 0 0 0s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=7d6b98897d
my-deployment-66494996ff 2 3 3 3m24s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
my-deployment-7d6b98897d 2 0 0 1s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=7d6b98897d
my-deployment-7d6b98897d 2 2 0 1s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=7d6b98897d
my-deployment-66494996ff 2 3 3 3m25s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
my-deployment-7d6b98897d 3 2 0 2s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=7d6b98897d
my-deployment-66494996ff 2 2 2 3m25s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
my-deployment-7d6b98897d 3 2 0 3s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=7d6b98897d
my-deployment-7d6b98897d 3 3 0 3s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=7d6b98897d
my-deployment-7d6b98897d 3 3 1 27s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=7d6b98897d
my-deployment-66494996ff 1 2 2 3m50s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
my-deployment-7d6b98897d 3 3 2 28s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=7d6b98897d
my-deployment-66494996ff 1 2 2 3m51s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
my-deployment-66494996ff 1 1 1 3m51s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
my-deployment-66494996ff 0 1 1 3m51s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
my-deployment-66494996ff 0 1 1 3m51s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
my-deployment-66494996ff 0 0 0 3m51s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=66494996ff
my-deployment-7d6b98897d 3 3 3 29s nginx-test nginx:stable-alpine-perl app=nginx,pod-template-hash=7d6b98897d
# deployment的replicas为3，根据升级策略，maxSurge=2，表示最多可以比replicas多两个，maxUnavailable=1，表示最多比replicas小一个。
# 从rs的pod数量变化也能看出来，升级一开始，新的rs就创建了两个pod，然后停掉了原来rs的一个pod。
# 回滚操作
[root@master-worker-node-1 deployment]# kubectl rollout undo deployment my-deployment --to-revision 1
小结
deployment是建立在replicaset之上的一个控制器，它拥有比rs更多的优点，滚动升级、升级策略多样、蓝绿发布、金丝雀发布、回滚方便
deployment的扩容缩容方法很多，只要能触发副本数量变化的方法都可以。比如：通过yaml文件apply、通过kubectl edit、通过kubectl scale deployment、通过kubectl patch等等
deployment的滚动升级可以配置相应的升级策略。升级的触发也可以通过yaml文件kubectl apply、也可以通过kubectl set image deployment/完成