SRE文档1.Deployment
1.0 Deployment资源定义规范
yaml
apiVersion: apps/v1 # API群组及版本
kind: Deployment # 资源类型特有标识
metadata:
name <string> # 资源名称，在作用域中要唯一
namespace <string> # 名称空间；Deployment隶属名称空间级别
spec:
minReadySeconds <integer> # Pod就绪后多少秒内任一容器无crash方可视为“就绪”
replicas <integer> # 期望的Pod副本数，默认为1
selector <object> # 标签选择器，必须匹配template字段中Pod模板中的标签
template <object> # Pod模板对象
revisionHistoryLimit <integer> # 滚动更新历史记录数量，默认为10
strategy <Object> # 滚动更新策略
type <string> # 滚动更新类型，可用值有Recreate和RollingUpdate；
rollingUpdate <Object> # 滚动更新参数，专用于RollingUpdate类型
maxSurge <string> # 更新期间可比期望的Pod数量多出的数量或比例；
maxUnavailable <string> # 更新期间可比期望的Pod数量缺少的数量或比例，10，
progressDeadlineSeconds <integer> # 滚动更新故障超时时长，默认为600秒
paused <boolean> # 是否暂停部署过程apiVersion: apps/v1 # API群组及版本
kind: Deployment # 资源类型特有标识
metadata:
name <string> # 资源名称，在作用域中要唯一
namespace <string> # 名称空间；Deployment隶属名称空间级别
spec:
minReadySeconds <integer> # Pod就绪后多少秒内任一容器无crash方可视为“就绪”
replicas <integer> # 期望的Pod副本数，默认为1
selector <object> # 标签选择器，必须匹配template字段中Pod模板中的标签
template <object> # Pod模板对象
revisionHistoryLimit <integer> # 滚动更新历史记录数量，默认为10
strategy <Object> # 滚动更新策略
type <string> # 滚动更新类型，可用值有Recreate和RollingUpdate；
rollingUpdate <Object> # 滚动更新参数，专用于RollingUpdate类型
maxSurge <string> # 更新期间可比期望的Pod数量多出的数量或比例；
maxUnavailable <string> # 更新期间可比期望的Pod数量缺少的数量或比例，10，
progressDeadlineSeconds <integer> # 滚动更新故障超时时长，默认为600秒
paused <boolean> # 是否暂停部署过程1.1 什么是deployment
DepLoyment（简称为depLoy）是Kubernetes控制器的一种高级别实现，他构建于RepLicaSet控制器之上。
我们只需要描述DepLoyment中的目标Pod期望状态，而DepLoyment控制器以受控速率更改实际状态，使其变为期望状态，也就是说，后期我们部署应用不直接使用Pod和RepLicaSet，而是使用DepLoyment控制器来调用RepLicaSet来实现，DepLoyment控制器在RepLicaSet原有基础上，添加了部分特性。
- 事件和状态查看：可以通过特定的命令查看Deployment对象的更新进度和状态
- 版本记录：将Deployment对象的更新操作都进行保存，以便后续执行回滚操作使用
- 多种更新方案：Recreate重建，可以实现单批次更新所有Pod。RoLLingUpdate可以实现多批次逐步替换Pod
1.2 deployment组成部分
Deployment 资源对象的格式和 ReplicaSet ⼏乎⼀致，Deployment 控制器也包含了3个基本的组成部分
- selector 标签选择器：匹配并关联Pod对象，并对授其管控的Pod对象计数；
- replicas期望的副本数：期望在集群中所运⾏的Pod对象数量
- template Pod模板：实际上就是定义的 Pod 内容，相当于把⼀个Pod 的描述以模板的形式嵌⼊到了 ReplicaSet
示例
- 编写资源配置清单， deployment-demo.yaml
yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: mall-deploy # 定义deploy资源名称
spec:
replicas: 2 # 定义deploy控制Pod的副本数
minReadySeconds: 10 # 定义Pod启动后，至少需要等待10秒钟，才能被认为是可用的
selector:
matchLabels: # 定义Pod选择器
app: mall
template: # Pod 模板，申明Pod名称、使用镜像
metadata:
labels: # 定义Pod标签
app: mall
spec:
containers:
- name: mall
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v1
ports:
- containerPort: 8080apiVersion: apps/v1
kind: Deployment
metadata:
name: mall-deploy # 定义deploy资源名称
spec:
replicas: 2 # 定义deploy控制Pod的副本数
minReadySeconds: 10 # 定义Pod启动后，至少需要等待10秒钟，才能被认为是可用的
selector:
matchLabels: # 定义Pod选择器
app: mall
template: # Pod 模板，申明Pod名称、使用镜像
metadata:
labels: # 定义Pod标签
app: mall
spec:
containers:
- name: mall
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v1
ports:
- containerPort: 80802.应用配置清单
bash
VERSION=v1 envsubst < deployment-demo.yaml | kubectl apply -f -VERSION=v1 envsubst < deployment-demo.yaml | kubectl apply -f -bash
#查看deployment信息
[root@kube-master yaml]# kubectl get deployments.apps
NAME READY UP-TO-DATE AVAILABLE AGE
deployment-demo 4/4 4 4 24s
#查看dp
[root@kube-master yaml]# kubectl get pods -l 'app=demoapp,release=stable'
NAME READY STATUS RESTARTS AGE
deployment-demo-54c55c8847-7qw8x 1/1 Running 0 116s
deployment-demo-54c55c8847-ch7hp 1/1 Running 0 116s
deployment-demo-54c55c8847-vq9l5 1/1 Running 0 115s
deployment-demo-54c55c8847-x99k6 1/1 Running 0 116s
# 可以看到第一段为deployment名字，最后一段为随机值，中间的fb544c5d8为replicaset中Pod模板的哈希值，也就是template字段的哈希值
# 查看replicaset信息
[root@kube-master yaml]# kubectl get rs
NAME DESIRED CURRENT READY AGE
deployment-demo-54c55c8847 4 4 4 5m40s
# 一旦pod模板发生变更，会导致ReplicaSet的哈希值发生变化，然后出发deployment更新的#查看deployment信息
[root@kube-master yaml]# kubectl get deployments.apps
NAME READY UP-TO-DATE AVAILABLE AGE
deployment-demo 4/4 4 4 24s
#查看dp
[root@kube-master yaml]# kubectl get pods -l 'app=demoapp,release=stable'
NAME READY STATUS RESTARTS AGE
deployment-demo-54c55c8847-7qw8x 1/1 Running 0 116s
deployment-demo-54c55c8847-ch7hp 1/1 Running 0 116s
deployment-demo-54c55c8847-vq9l5 1/1 Running 0 115s
deployment-demo-54c55c8847-x99k6 1/1 Running 0 116s
# 可以看到第一段为deployment名字，最后一段为随机值，中间的fb544c5d8为replicaset中Pod模板的哈希值，也就是template字段的哈希值
# 查看replicaset信息
[root@kube-master yaml]# kubectl get rs
NAME DESIRED CURRENT READY AGE
deployment-demo-54c55c8847 4 4 4 5m40s
# 一旦pod模板发生变更，会导致ReplicaSet的哈希值发生变化，然后出发deployment更新的3.对deployment实行滚动更新
bash
VERSION=v2 envsubst < deployment-demo.yaml | kubectl apply -f -
#使用rollout查看滚动更新的状态
[root@kube-master yaml]# kubectl rollout status deployment deployment-demo
deployment "deployment-demo" successfully rolled out
#查看pod是否更新为v1.2版本
[root@kube-master yaml]# curl 172.30.0.181
hsuing demoapp v1.2 !! ClientIP: 172.25.244.192, PodName: deployment-demo-74b48bd6db-kd2j2, PodIP: 172.30.0.181!VERSION=v2 envsubst < deployment-demo.yaml | kubectl apply -f -
#使用rollout查看滚动更新的状态
[root@kube-master yaml]# kubectl rollout status deployment deployment-demo
deployment "deployment-demo" successfully rolled out
#查看pod是否更新为v1.2版本
[root@kube-master yaml]# curl 172.30.0.181
hsuing demoapp v1.2 !! ClientIP: 172.25.244.192, PodName: deployment-demo-74b48bd6db-kd2j2, PodIP: 172.30.0.181!4.回滚
bash
#查看deployment的更新历史信息
[root@kube-master yaml]# kubectl rollout history deployment deployment-demo
deployment.apps/deployment-demo
REVISION CHANGE-CAUSE
1 <none>
2 <none> # 此为当前版本信息
#将其回滚到上一个版本
[root@kube-master yaml]# kubectl rollout undo deployment deployment-demo
deployment.apps/deployment-demo rolled back
#查看replicaset信息
[root@kube-master yaml]# kubectl get rs
NAME DESIRED CURRENT READY AGE
deployment-demo-54c55c8847 4 4 4 13m
deployment-demo-74b48bd6db 0 0 0 6m15s
#再次查看更新历史
[root@kube-master yaml]# kubectl rollout history deployment deployment-demo
deployment.apps/deployment-demo
REVISION CHANGE-CAUSE
2 <none>
3 <none>
# 可以看到原本的1没有了新增了3，3表示当前的#查看deployment的更新历史信息
[root@kube-master yaml]# kubectl rollout history deployment deployment-demo
deployment.apps/deployment-demo
REVISION CHANGE-CAUSE
1 <none>
2 <none> # 此为当前版本信息
#将其回滚到上一个版本
[root@kube-master yaml]# kubectl rollout undo deployment deployment-demo
deployment.apps/deployment-demo rolled back
#查看replicaset信息
[root@kube-master yaml]# kubectl get rs
NAME DESIRED CURRENT READY AGE
deployment-demo-54c55c8847 4 4 4 13m
deployment-demo-74b48bd6db 0 0 0 6m15s
#再次查看更新历史
[root@kube-master yaml]# kubectl rollout history deployment deployment-demo
deployment.apps/deployment-demo
REVISION CHANGE-CAUSE
2 <none>
3 <none>
# 可以看到原本的1没有了新增了3，3表示当前的2.Deployment重建策略
2.1 什么是Recreate
重建（Recreate）当更新策略设定为Recreate，在更新镜像时，它会先杀死正在运行的Pod，等彻底杀死后，重新创建新的RS，然后启动对应的Pod，那么在这个更新过程中，会造成服务一段时间无法提供服务；
第一步：同时杀死所有旧版本的Pod，此时Pod无法正常对外提供服务；）
第二步：创建新的RS，启动新的Pod;
第三步：等待Pod就绪，对外提供服务；
❌ 注意
通常只有当应用的新旧版本不兼容（例如依赖的后端数据的格式不同且无法兼容）时才会使用Recreate重建策略。
2.2 Recreate实践
yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: demo-deployment
spec:
replicas: 3
strategy:
type: Recreate
selector:
matchLabels:
app: demo
template:
metadata:
labels:
app: demo
spec:
containers:
- name: container-recreate #定义名字不能有下划线
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v2
ports:
- containerPort: 80apiVersion: apps/v1
kind: Deployment
metadata:
name: demo-deployment
spec:
replicas: 3
strategy:
type: Recreate
selector:
matchLabels:
app: demo
template:
metadata:
labels:
app: demo
spec:
containers:
- name: container-recreate #定义名字不能有下划线
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v2
ports:
- containerPort: 80- 观察现象
shell
[root@kube-master yaml]# kubectl apply -f demoapp_recreate.yaml
[root@kube-master yaml]# kubectl get pod
NAME READY STATUS RESTARTS AGE
demo-deployment-bf8f88dbc-4d298 1/1 Terminating 3 (7h45m ago) 10d
demo-deployment-bf8f88dbc-b5dmc 1/1 Terminating 3 (7h45m ago) 10d
demo-deployment-bf8f88dbc-tfl2g 1/1 Terminating 3 (7h45m ago) 10d
demo-deployment-bf8f88dbc-xrjfm 1/1 Terminating 3 (7h45m ago) 9d
[root@kube-master yaml]# kubectl get pod
NAME READY STATUS RESTARTS AGE
demo-deployment-5684f48684-7zx87 1/1 Running 0 87s
demo-deployment-5684f48684-nlwhp 1/1 Running 0 87s
demo-deployment-5684f48684-zz6kc 1/1 Running 0 87s
[root@kube-master yaml]# kubectl get rs
NAME DESIRED CURRENT READY AGE
demo-deployment-5684f48684 3 3 3 108s
demo-deployment-bf8f88dbc 0 0 0 10d[root@kube-master yaml]# kubectl apply -f demoapp_recreate.yaml
[root@kube-master yaml]# kubectl get pod
NAME READY STATUS RESTARTS AGE
demo-deployment-bf8f88dbc-4d298 1/1 Terminating 3 (7h45m ago) 10d
demo-deployment-bf8f88dbc-b5dmc 1/1 Terminating 3 (7h45m ago) 10d
demo-deployment-bf8f88dbc-tfl2g 1/1 Terminating 3 (7h45m ago) 10d
demo-deployment-bf8f88dbc-xrjfm 1/1 Terminating 3 (7h45m ago) 9d
[root@kube-master yaml]# kubectl get pod
NAME READY STATUS RESTARTS AGE
demo-deployment-5684f48684-7zx87 1/1 Running 0 87s
demo-deployment-5684f48684-nlwhp 1/1 Running 0 87s
demo-deployment-5684f48684-zz6kc 1/1 Running 0 87s
[root@kube-master yaml]# kubectl get rs
NAME DESIRED CURRENT READY AGE
demo-deployment-5684f48684 3 3 3 108s
demo-deployment-bf8f88dbc 0 0 0 10d3.Deployment滚动更新
https://kubernetes.io/zh-cn/docs/tutorials/kubernetes-basics/update/update-intro/
shell
kubectl explain deploy.spec.strategykubectl explain deploy.spec.strategy3.1 什么是滚动更新
滚动更新（RoLLingUpdate），一次仅更新一批Pod，当更新的Pod就绪后，在更新另一批，直到全部更新完成为止；该策略实现了不间断服务的目标，在更新过程中可能会出现不同的应用版本并存且，同时提供服务的情况。
- 第一步：创建新的RepLicaSet，然后根据新的镜像运行新的Pod;
- 第二步：删除l旧的Pod，启动新的Pod，新Pod就绪后，继续删除旧Pod，启动新Pod;
- 第三步：持续第二步过程，一直到所有Pod都被更新成功。
3.2 实践
yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: demo-deployment
spec:
replicas: 3
strategy:
type: RollingUpdate
selector:
matchLabels:
app: demo
template:
metadata:
labels:
app: demo
spec:
containers:
- name: container-recreate
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v2
ports:
- containerPort: 80apiVersion: apps/v1
kind: Deployment
metadata:
name: demo-deployment
spec:
replicas: 3
strategy:
type: RollingUpdate
selector:
matchLabels:
app: demo
template:
metadata:
labels:
app: demo
spec:
containers:
- name: container-recreate
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v2
ports:
- containerPort: 80- 应用
shell
[root@kube-master yaml]# kubectl apply -f demoapp_RollingUpdate.yaml
deployment.apps/demo-deployment configured[root@kube-master yaml]# kubectl apply -f demoapp_RollingUpdate.yaml
deployment.apps/demo-deployment configured- 观察
shell
[root@kube-master yaml]# kubectl describe deployments.apps demo-deployment
...
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal ScalingReplicaSet 5m33s deployment-controller Scaled down replica set demo-deployment-5684f48684 to 0
Normal ScalingReplicaSet 4m59s deployment-controller Scaled up replica set demo-deployment-645bbb7877 to 3
Normal ScalingReplicaSet 63s deployment-controller Scaled up replica set demo-deployment-5684f48684 to 1
Normal ScalingReplicaSet 60s deployment-controller Scaled down replica set demo-deployment-645bbb7877 to 2
Normal ScalingReplicaSet 60s deployment-controller Scaled up replica set demo-deployment-5684f48684 to 2
Normal ScalingReplicaSet 58s deployment-controller Scaled down replica set demo-deployment-645bbb7877 to 1
Normal ScalingReplicaSet 58s deployment-controller Scaled up replica set demo-deployment-5684f48684 to 3
Normal ScalingReplicaSet 56s deployment-controller Scaled down replica set demo-deployment-645bbb7877 to 0[root@kube-master yaml]# kubectl describe deployments.apps demo-deployment
...
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal ScalingReplicaSet 5m33s deployment-controller Scaled down replica set demo-deployment-5684f48684 to 0
Normal ScalingReplicaSet 4m59s deployment-controller Scaled up replica set demo-deployment-645bbb7877 to 3
Normal ScalingReplicaSet 63s deployment-controller Scaled up replica set demo-deployment-5684f48684 to 1
Normal ScalingReplicaSet 60s deployment-controller Scaled down replica set demo-deployment-645bbb7877 to 2
Normal ScalingReplicaSet 60s deployment-controller Scaled up replica set demo-deployment-5684f48684 to 2
Normal ScalingReplicaSet 58s deployment-controller Scaled down replica set demo-deployment-645bbb7877 to 1
Normal ScalingReplicaSet 58s deployment-controller Scaled up replica set demo-deployment-5684f48684 to 3
Normal ScalingReplicaSet 56s deployment-controller Scaled down replica set demo-deployment-645bbb7877 to 0❌ 注意
会发现新老版本存在交替，但并不会出现服务不可用状态；
3.3 应⽤回退实践
有时候，你可能想要回滚 DepLoyment；例如，当DepLoyment 不稳定进入反复崩溃状态。默认情况下，DepLoyment的所有上线记录都保留在系统中，以便可以随时回滚。（你可以通过修改revisionHistoryLimit调整保留的数量，默认10条)
检查 Deployment 上线的历史版本
shell
[root@kube-master yaml]# kubectl rollout history deployment <deployment_name>
deployment.apps/demo-deployment
REVISION CHANGE-CAUSE
1 <none>
3 <none>
4 <none>[root@kube-master yaml]# kubectl rollout history deployment <deployment_name>
deployment.apps/demo-deployment
REVISION CHANGE-CAUSE
1 <none>
3 <none>
4 <none>查看revision对应版本
shell
[root@kube-master yaml]# kubectl rollout history deployment demo-deployment --revision=4
deployment.apps/demo-deployment with revision #4
Pod Template:
Labels: app=demo
pod-template-hash=645bbb7877
Containers:
container-recreate:
Image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v1
Port: 80/TCP
Host Port: 0/TCP
Environment: <none>
Mounts: <none>
Volumes: <none>[root@kube-master yaml]# kubectl rollout history deployment demo-deployment --revision=4
deployment.apps/demo-deployment with revision #4
Pod Template:
Labels: app=demo
pod-template-hash=645bbb7877
Containers:
container-recreate:
Image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v1
Port: 80/TCP
Host Port: 0/TCP
Environment: <none>
Mounts: <none>
Volumes: <none>开始回退
通过--to-revision参数回退
shell
[root@kube-master yaml]# kubectl rollout undo deployment demo-deployment --to-revision=4
deployment.apps/demo-deployment rolled back[root@kube-master yaml]# kubectl rollout undo deployment demo-deployment --to-revision=4
deployment.apps/demo-deployment rolled back4.Deployment更新策略
shell
kubectl explain deploy.spec.strategykubectl explain deploy.spec.strategyDeployment 会在 .spec.strategy.type=RollingUpdate时，采取滚动更新的方式更新 Pods。可以指定maxUnavailable和maxSurgeS来控制滚动更新过程
maxSurge最大可用Pod
- 用来指定可以创建超出期望Pod个数的Pod数量。可以是数字，也可以是百分比（例如10%）此字段的默认值为25%
- 例如，当此值为20%时，启动滚动更新后，会立即对新的ReplicaSet扩容，同时保证新l旧 Pod 的总数不超过所需Pod 总数的 120%。一旦l旧 Pods 被杀死，新的ReplicaSet 可以进一步扩容，同时确保更新期间的任何时候运行中的 Pods 总数最多为所需 Pods 总数的 120%。计算公式：
10+（10x20%）=12
maxUnavailable最大不可用Pod
- 用来指定更新过程中不可用的 Pod 的个数上限。可以是数字,也可以是百分比 (例如10%)此字段的默认值为25%
- 例如，当此值设置为20%时，滚动更新开始时会立即将旧RepLicaSet 缩容到期望 Pod 个数的7θ%。新Pod 准备就绪后，继续缩容l旧有的 ReplicaSet，然后对新的ReplicaSet 扩容，确保在更新期间可用的 Pods总数在任何时候都是所需的 Pod 个数的 70%。计算公式：
10-(10x20%)=8
❌ 注意
maxSurge和maxUnavaiLable两个属性协同工作，可组合定义出3中不同的策略完成多批次的应用更新。
1-先增新，后减旧：将maxSurge设置为30%，将maxUnavailable的值设为θ;
2-先减旧，后增新：将maxUnavaiLabLe设置为30%，将maxSurge的值设为θ;
3-同时增减，将maxSurge和maxUnavaiLabLe分别设定为20%;期望是12Pod，至少就绪8个Pod
4.1 maxSurge
指定升级期间存在的总Pod对象数量最多可超出期望值的个数，可以是0，也可以是整数，也可以是⼀个百分⽐
例如：期望的的值为10，maxSurge属性为2，则表示Pod对象总数不能超过12个。计算公式：10+（10x20%）=12
实践
shell
[root@kube-master yaml]# cat demoapp_RollingUpdate_max.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: demo-deployment
spec:
replicas: 10
strategy:
rollingUpdate:
maxSurge: 20% # 最⼤可⽤20%
maxUnavailable: 0 # 不可⽤状态设定为0
selector:
matchLabels:
app: demo
template:
metadata:
labels:
app: demo
spec:
containers:
- name: container-recreate
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v2
ports:
- containerPort: 80[root@kube-master yaml]# cat demoapp_RollingUpdate_max.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: demo-deployment
spec:
replicas: 10
strategy:
rollingUpdate:
maxSurge: 20% # 最⼤可⽤20%
maxUnavailable: 0 # 不可⽤状态设定为0
selector:
matchLabels:
app: demo
template:
metadata:
labels:
app: demo
spec:
containers:
- name: container-recreate
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v2
ports:
- containerPort: 80- 查看
shell
watch -n 1 kubectl get rs
kubectl describe deployments.apps demo-deploymentwatch -n 1 kubectl get rs
kubectl describe deployments.apps demo-deployment4.2 maxUnavailable
升级期间不可⽤的Pod副本数（包括新版本），最多不能低于期望的个数，默认为1
如：期望的值为1θ，maxunavaiabLe属性为2，则表示Pod处于正常状态至少有8个。计算公式：: 10-（10x20%)=8
实践
yaml
[root@kube-master yaml]# cat demoapp_RollingUpdate_max.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: demo-deployment
spec:
replicas: 5
strategy:
rollingUpdate:
maxSurge: 0 # 最⼤可⽤0
maxUnavailable: 20% # 不可⽤Pod副本最低20%
selector:
matchLabels:
app: demo
template:
metadata:
labels:
app: demo
spec:
containers:
- name: container-recreate
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v2
ports:
- containerPort: 80[root@kube-master yaml]# cat demoapp_RollingUpdate_max.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: demo-deployment
spec:
replicas: 5
strategy:
rollingUpdate:
maxSurge: 0 # 最⼤可⽤0
maxUnavailable: 20% # 不可⽤Pod副本最低20%
selector:
matchLabels:
app: demo
template:
metadata:
labels:
app: demo
spec:
containers:
- name: container-recreate
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v2
ports:
- containerPort: 80- 观察
shell
watch -n 1 kubectl get rswatch -n 1 kubectl get rs4.3 maxSurge & MaxUnavailable
同时设定maxSurge 以及 MaxUnavailable
yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: demo-deployment
spec:
replicas: 5
strategy:
rollingUpdate:
maxSurge: 20% # 最⼤可⽤0
maxUnavailable: 20% # 不可⽤Pod副本最低20%
selector:
matchLabels:
app: demo
template:
metadata:
labels:
app: demo
spec:
containers:
- name: container-recreate
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v2
ports:
- containerPort: 80apiVersion: apps/v1
kind: Deployment
metadata:
name: demo-deployment
spec:
replicas: 5
strategy:
rollingUpdate:
maxSurge: 20% # 最⼤可⽤0
maxUnavailable: 20% # 不可⽤Pod副本最低20%
selector:
matchLabels:
app: demo
template:
metadata:
labels:
app: demo
spec:
containers:
- name: container-recreate
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v2
ports:
- containerPort: 804.4 minReadySeconds
Deployment支持使用spec.minReadySeconds字段来控制滚动更新的速度，默认值为θ，表示新建的Pod对象一旦“就绪"将立即被视作可用，随后即可开始下一轮更新过程。如果设定了spec.minReadySeconds：3及表示新建的Pod对象至少要成功运行多久才会被视作可用，即就绪之后还要等待指定的3s才能开始下一批次的更新。在一个批次内新建的所有Pod就绪后在转为可用状态前，更新操作会被阻塞，并且任何一个Pod就绪探测失败，都会导致滚动更新被终止。
因此，为minReadySeconds设定一个合理的值，不仅能够减缓更新的速度，还能够让DepLoyment提前发现一部分程序因为Bug导致的升级故障。
4.5 revisionHistoryLimit
Deployment保留一部分更新历史中旧版本的ReplicaSet对象，当我们执行回滚操作的时候，就直接使用旧版本的ReplicaSet，在Deployment资源保存历史版本数量有spec.revisionHistoryLimit 属性进行定义
shell
[root@kube-master yaml]# kubectl rollout history deployment
deployment.apps/demo-deployment
REVISION CHANGE-CAUSE
2 <none>
3 <none>
deployment.apps/nginx
REVISION CHANGE-CAUSE
1 <none>[root@kube-master yaml]# kubectl rollout history deployment
deployment.apps/demo-deployment
REVISION CHANGE-CAUSE
2 <none>
3 <none>
deployment.apps/nginx
REVISION CHANGE-CAUSE
1 <none>4.6 progressDeadlineSeconds
滚动更新故障超时时长，默认为600秒，k8s在升级过程中有可能由于各种原因升级卡住（这个时候还没有明确的升级失败），比如在拉取被墙的镜像，权限不够等错误。如果配置progressDeadlineSeconds，当达到了时间如果还卡着，则会上报这个异常情况，这个时候这个Deployment状态就被标记为False，并且注明原因。但是它并不会阻止Deployment继续进行卡住后面的升级操作.
4.7 红绿发布
1.0
yaml
[root@kube-master yaml]# cat hong_lu1.0.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: demoapp-version-10-prod
spec:
replicas: 3
selector:
matchLabels:
app: demoapp
version: v1.0
template:
metadata:
labels:
app: demoapp
version: v1.0
spec:
containers:
- name: demoapp
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v1.0
---
apiVersion: v1
kind: Service
metadata:
name: demoapp-version-service
spec:
selector:
app: demoapp
ports:
- name: http
port: 80
targetPort: 80[root@kube-master yaml]# cat hong_lu1.0.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: demoapp-version-10-prod
spec:
replicas: 3
selector:
matchLabels:
app: demoapp
version: v1.0
template:
metadata:
labels:
app: demoapp
version: v1.0
spec:
containers:
- name: demoapp
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v1.0
---
apiVersion: v1
kind: Service
metadata:
name: demoapp-version-service
spec:
selector:
app: demoapp
ports:
- name: http
port: 80
targetPort: 801.1
yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: demoapp-version-11-prod
spec:
replicas: 1
selector:
matchLabels:
app: demoapp
version: v1.1
template:
metadata:
labels:
app: demoapp
version: v1.1
spec:
containers:
- name: demoapp
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v1.1apiVersion: apps/v1
kind: Deployment
metadata:
name: demoapp-version-11-prod
spec:
replicas: 1
selector:
matchLabels:
app: demoapp
version: v1.1
template:
metadata:
labels:
app: demoapp
version: v1.1
spec:
containers:
- name: demoapp
image: registry.cn-zhangjiakou.aliyuncs.com/hsuing/demoapp:v1.1- 访问clusterIP
shell
kubectl get svc
curl clusterIp/versionkubectl get svc
curl clusterIp/version