Kubernetes-Deployment

2023-10-18

字数统计: 2.7k字 | 阅读时长≈ 13分

Deployment

Kubernetes 提供了一种更加简单的更新 RC 和 Pod 的机制，叫做 Deployment。通过在 Deployment 中描述期望的集群状态，Deployment Controller 会将现在的集群状态在一个可控的速度下逐步更新成期望的集群状态。Deployment 过程如下图所示：

Deployment 的主要职责同样是为了保证 Pod 的数量和健康，而绝大多数的功能与 Replication Controller 完全一样，因此可以被看作新一代的 Replication Controller。但是，它又具备了 Replication Controller 不具备的新特性：

事件和状态查看：可以查看升级的详细进度和状态；
回滚：当升级 Pod 镜像或者相关参数的时候发现问题，可以使用回滚操作回滚到上一个稳定的版本或者指定的版本；
版本记录：每一次对 Deployment 的操作都能够保存下来，给与后续可能的回滚使用；
暂停和启动：对于每一次升级，都能够随时暂停和启动；
多种升级方案：
- Recreate: 删除所有已存在的 Pod，重新创建新的；
- RollingUpdate: 滚动升级，即逐步替换的策略。滚动升级时支持更多的附件参数，例如设置最大不可用 Pod 数量，最小升级间隔时间等。

相比于 RC，Deployment 直接使用 kubectl edit deployment/deploymentName 或者 kubectl set 方法就可以直接升级（原理是 Pod 的 Template 发生变化，例如，更新 Label，更新镜像版本等操作会触发 Deployment 的滚动升级）。

创建 Deployment

首先，我们定义一个 nginx-deploy-v1.yaml 文件，副本数为 3，如下：

apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-deployment
  labels:
    app: nginx
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx:1.7.9
        ports:
        - containerPort: 80

示例解析

第4行，metadata.name: Deployment 的名称
第8行，spec.replicas: 创建 Pod 的副本数
第9行，spec.selector: 定义 Deployment 如何找到要管理的 Pod，与 template 下的 labels 对应
template 字段包含以下字段：
- metadata.labels.app: 使用 label 标记 Pod
- spec.containers.name: 表示 Pod 运行一个名字为 nginx 的容器
- spec.containers.image: 运行此 Pod 使用的镜像
- Ports: 容器用于发送和接收流量的端口。

使用 kubectl create 创建 Deployment

1 2	# kubectl create -f nginx-deploy-v1.yaml deployment.apps/nginx-deployment created

使用 kubectl get 或者 kubectl describe 查看此 Deployment

1
2
3

kubectl get deploy
NAME               READY   UP-TO-DATE   AVAILABLE   AGE
nginx-deployment   3/3     3            3           56s

此时，我们查看 Deployment rollout 的状态

1 2	# kubectl rollout status deployment nginx-deployment deployment "nginx-deployment" successfully rolled out

查看此 Deployment 创建的 ReplicaSet

1
2
3

# kubectl get rs
NAME                          DESIRED   CURRENT   READY   AGE
nginx-deployment-5d59d67564   3         3         3       15m

查看此 Deployment 创建的 Pod

# kubectl get pods --show-labels 
NAME                                READY   STATUS    RESTARTS   AGE   LABELS
nginx-deployment-5d59d67564-2dv47   1/1     Running   0          21m   app=nginx,pod-template-hash=5d59d67564
nginx-deployment-5d59d67564-lrhhp   1/1     Running   0          21m   app=nginx,pod-template-hash=5d59d67564
nginx-deployment-5d59d67564-mdnf5   1/1     Running   0          21m   app=nginx,pod-template-hash=5d59d67564

更新 Deployment

一般应用程序升级或者迭代时，会通过 Deployment 对 Pod 进行滚动更新。

假如更新 Nginx Pod 的 image 使用 nginx:1.9.1

1 2	# kubectl set image deploy nginx-deployment nginx=nginx:1.9.1 --record deployment.apps/nginx-deployment image updated

也可以使用 edit 直接编辑 Deployment，效果相同
1
kubectl edit deployment nginx-deployment

使用 kubectl rollout status 查看更新状态

# kubectl rollout status deployment nginx-deployment 
Waiting for deployment "nginx-deployment" rollout to finish: 2 out of 3 new replicas have been updated...
Waiting for deployment "nginx-deployment" rollout to finish: 2 out of 3 new replicas have been updated...
Waiting for deployment "nginx-deployment" rollout to finish: 1 old replicas are pending termination...
Waiting for deployment "nginx-deployment" rollout to finish: 1 old replicas are pending termination...
deployment "nginx-deployment" successfully rolled out

查看 ReplicaSet 状态

# kubectl get rs
NAME                          DESIRED   CURRENT   READY   AGE
nginx-deployment-5d59d67564   0         0         0       32m
nginx-deployment-69c44dfb78   3         3         3       5m3s

通过 describe 查看 deployment 的详细信息

# kubectl describe deployments nginx-deployment 
Name:                   nginx-deployment
Namespace:              default
CreationTimestamp:      Sat, 03 Apr 2021 21:14:15 +0800
Labels:                 app=nginx
Annotations:            deployment.kubernetes.io/revision: 4
                        kubernetes.io/change-cause: kubectl set image deploy nginx-deployment nginx=nginx:1.9.1 --record=true
Selector:               app=nginx
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=nginx
Containers:
nginx:
    Image:        nginx:1.9.1
    Port:         80/TCP
    Host Port:    0/TCP
    Environment:  <none>
    Mounts:       <none>
Volumes:        <none>
Conditions:
Type           Status  Reason
----           ------  ------
Available      True    MinimumReplicasAvailable
Progressing    True    NewReplicaSetAvailable
OldReplicaSets:  <none>
NewReplicaSet:   nginx-deployment-69c44dfb78 (3/3 replicas created)
Events:
Type    Reason             Age                    From                   Message
----    ------             ----                   ----                   -------
Normal  ScalingReplicaSet  33m                    deployment-controller  Scaled up replica set nginx-deployment-5d59d67564 to 3
Normal  ScalingReplicaSet  6m                     deployment-controller  Scaled up replica set nginx-deployment-69c44dfb78 to 1
Normal  ScalingReplicaSet  5m16s                  deployment-controller  Scaled down replica set nginx-deployment-5d59d67564 to 2
Normal  ScalingReplicaSet  4m31s                  deployment-controller  Scaled up replica set nginx-deployment-69c44dfb78 to 3
Normal  ScalingReplicaSet  2m25s                  deployment-controller  Scaled up replica set nginx-deployment-5d59d67564 to 1
Normal  ScalingReplicaSet  2m24s                  deployment-controller  Scaled up replica set nginx-deployment-5d59d67564 to 2
Normal  ScalingReplicaSet  2m24s                  deployment-controller  Scaled down replica set nginx-deployment-69c44dfb78 to 2
Normal  ScalingReplicaSet  2m11s (x2 over 5m16s)  deployment-controller  Scaled up replica set nginx-deployment-69c44dfb78 to 2
Normal  ScalingReplicaSet  2m10s (x2 over 4m31s)  deployment-controller  Scaled down replica set nginx-deployment-5d59d67564 to 1
Normal  ScalingReplicaSet  2m10s (x6 over 2m23s)  deployment-controller  (combined from similar events): Scaled up replica set nginx-deployment-69c44dfb78 to 3
Normal  ScalingReplicaSet  102s (x2 over 4m3s)    deployment-controller  Scaled down replica set nginx-deployment-5d59d67564 to 0

从 describe 中可以看出，第一次创建时，它创建了一个名为 nginx-deployment-5d59d67564 的 ReplicaSet，并直接将其扩展为 3 个副本。更新部署时，它创建了一个新的 ReplicaSet，命名为 nginx-deployment-69c44dfb78，并将其副本数扩展为 1，然后将旧的的 ReplicaSet 缩小为 2，这样至少可以有 2 个 Pod 可用，最多创建了 4 个 Pod。以此类推，使用相同的滚动更新策略向上和向下扩展新旧 ReplicaSet，最终新的 ReplicaSet 可以拥有 3 个副本，并将旧的 ReplicaSet 缩小为 0.

回滚 Deployment

当心版本不稳定时，可以对其进行回滚操作，默认情况下，所有 Deployment 的 rollout 历史记录都保留在系统中，可以随时回滚。

假设我们又进行了几次更新

# kubectl set image deployment nginx-deployment nginx=dotbalo/canary:v1 --record
deployment.apps/nginx-deployment image updated

# kubectl set image deployment nginx-deployment nginx=dotbalo/canary:v2 --record
deployment.apps/nginx-deployment image updated

使用 kubectl rollout history 查看部署历史

# kubectl rollout history deployment nginx-deployment 
deployment.apps/nginx-deployment 
REVISION  CHANGE-CAUSE
3         kubectl set image deploy nginx-deployment nginx=nginx:1.9.1 --record=true
4         kubectl set image deploy nginx-deployment nginx=nginx:1.9.1 --record=true
5         kubectl set image deployment nginx-deployment nginx=dotbalo/canary:v1 --record=true
6         kubectl set image deployment nginx-deployment nginx=dotbalo/canary:v2 --record=true

查看 deployment 某次更新的详细信息，使用 --revision 指定版本号

# kubectl rollout history deployment nginx-deployment --revision=5
deployment.apps/nginx-deployment with revision #5
Pod Template:
Labels:       app=nginx
        pod-template-hash=666f99dd56
Annotations:  kubernetes.io/change-cause: kubectl set image deployment nginx-deployment nginx=dotbalo/canary:v1 --record=true
Containers:
nginx:
    Image:      dotbalo/canary:v1
    Port:       80/TCP
    Host Port:  0/TCP
    Environment:        <none>
    Mounts:     <none>
Volumes:      <none>

使用 kubectl rollout undo 回滚到上一个版本

1 2	# kubectl rollout undo deployment nginx-deployment deployment.apps/nginx-deployment rolled back

再次查看历史记录，发现 REVISION 7 回到了 canary:v1

# kubectl rollout history deployment nginx-deployment 
deployment.apps/nginx-deployment 
REVISION  CHANGE-CAUSE
3         kubectl set image deploy nginx-deployment nginx=nginx:1.9.1 --record=true
4         kubectl set image deploy nginx-deployment nginx=nginx:1.9.1 --record=true
6         kubectl set image deployment nginx-deployment nginx=dotbalo/canary:v2 --record=true
7         kubectl set image deployment nginx-deployment nginx=dotbalo/canary:v1 --record=true

使用 --to-revision 参数回滚到指定版本

1 2	# kubectl rollout undo deployment nginx-deployment --to-revision=3 deployment.apps/nginx-deployment rolled back

扩容/缩容 Deployment

当公司访问量变大，三个 Pod 已经无法支撑业务时，可以对其进行扩展。

使用 kubectl scale 动态调整 Pod 的副本数，比如增加 Pod 为 5 个

1 2	# kubectl scale deployment nginx-deployment --replicas=5 deployment.apps/nginx-deployment scaled

查看 Pod，此时 Pod 已经变成了 5 个

# kubectl get pods --show-labels 
NAME                                READY   STATUS    RESTARTS   AGE     LABELS
nginx-deployment-5d59d67564-99l2x   1/1     Running   0          3m20s   app=nginx,pod-template-hash=5d59d67564
nginx-deployment-5d59d67564-fl5q7   1/1     Running   0          71s     app=nginx,pod-template-hash=5d59d67564
nginx-deployment-5d59d67564-js2xn   1/1     Running   0          3m24s   app=nginx,pod-template-hash=5d59d67564
nginx-deployment-5d59d67564-sbfxw   1/1     Running   0          71s     app=nginx,pod-template-hash=5d59d67564
nginx-deployment-5d59d67564-vsl95   1/1     Running   0          3m22s   app=nginx,pod-template-hash=5d59d67564

暂停和恢复 Deployment 更新

Deployment 支持暂停更新，用于对 Deployment 进行多次修改操作。

使用 kubectl rollout pause 暂停 Deployment 更新

1 2	# kubectl rollout pause deployment nginx-deployment deployment.apps/nginx-deployment paused

对 Deployment 进行相关更新操作，比如更新镜像，然后对其资源进行限制

# kubectl set image deploy nginx-deployment nginx=nginx:1.14.2
deployment.apps/nginx-deployment image updated

# kubectl set resources deploy nginx-deployment -c=nginx --limits=cpu=200m,memory=254Mi
deployment.apps/nginx-deployment resource requirements updated

通过 rollout history 可以看到没有新的更新

# kubectl rollout history deployment nginx-deployment 
deployment.apps/nginx-deployment 
REVISION  CHANGE-CAUSE
4         kubectl set image deploy nginx-deployment nginx=nginx:1.9.1 --record=true
6         kubectl set image deployment nginx-deployment nginx=dotbalo/canary:v2 --record=true
7         kubectl set image deployment nginx-deployment nginx=dotbalo/canary:v1 --record=true
8         kubectl set image deploy nginx-deployment nginx=nginx:1.9.1 --record=true

使用 kubectl rollout resume 恢复 Deployment 的更新

1 2	# kubectl rollout resume deployment nginx-deployment deployment.apps/nginx-deployment resumed

可以看到恢复更新的 Deployment 创建了一个新的 RS（复制集）

# kubectl get rs
NAME                          DESIRED   CURRENT   READY   AGE
nginx-deployment-5b99ddccc6   5         5         5       54s
nginx-deployment-5d59d67564   0         0         0       59m
nginx-deployment-666f99dd56   0         0         0       19m
nginx-deployment-69c44dfb78   0         0         0       31m
nginx-deployment-79d4d8c7d5   0         0         0       18m

可以看到 Deployment 的 image 镜像已经变成了 nginx:1.14.2

# kubectl describe deployments.apps nginx-deployment 
Name:                   nginx-deployment
Namespace:              default
CreationTimestamp:      Sat, 03 Apr 2021 21:14:15 +0800
Labels:                 app=nginx
Annotations:            deployment.kubernetes.io/revision: 9
                        kubernetes.io/change-cause: kubectl set image deploy nginx-deployment nginx=nginx:1.9.1 --record=true
Selector:               app=nginx
Replicas:               5 desired | 5 updated | 5 total | 5 available | 0 unavailable
StrategyType:           RollingUpdate
MinReadySeconds:        0
RollingUpdateStrategy:  25% max unavailable, 25% max surge
Pod Template:
Labels:  app=nginx
Containers:
nginx:
    Image:      nginx:1.14.2
    Port:       80/TCP
    Host Port:  0/TCP
    Limits:
    cpu:        200m
    memory:     254Mi
    Environment:  <none>
    Mounts:       <none>
Volumes:        <none>
Conditions:
Type           Status  Reason
----           ------  ------
Available      True    MinimumReplicasAvailable
Progressing    True    NewReplicaSetAvailable
OldReplicaSets:  <none>
NewReplicaSet:   nginx-deployment-5b99ddccc6 (5/5 replicas created)
Events:
Type    Reason             Age                   From                   Message
----    ------             ----                  ----                   -------
Normal  ScalingReplicaSet  32m                   deployment-controller  Scaled up replica set nginx-deployment-69c44dfb78 to 1
Normal  ScalingReplicaSet  32m                   deployment-controller  Scaled down replica set nginx-deployment-5d59d67564 to 2
Normal  ScalingReplicaSet  31m                   deployment-controller  Scaled up replica set nginx-deployment-69c44dfb78 to 3
Normal  ScalingReplicaSet  29m                   deployment-controller  Scaled up replica set nginx-deployment-5d59d67564 to 1
Normal  ScalingReplicaSet  29m                   deployment-controller  Scaled up replica set nginx-deployment-5d59d67564 to 2
Normal  ScalingReplicaSet  29m (x2 over 32m)     deployment-controller  Scaled up replica set nginx-deployment-69c44dfb78 to 2
Normal  ScalingReplicaSet  29m (x2 over 31m)     deployment-controller  Scaled down replica set nginx-deployment-5d59d67564 to 1
Normal  ScalingReplicaSet  28m (x2 over 30m)     deployment-controller  Scaled down replica set nginx-deployment-5d59d67564 to 0
Normal  ScalingReplicaSet  18m (x2 over 29m)     deployment-controller  Scaled down replica set nginx-deployment-69c44dfb78 to 2
Normal  ScalingReplicaSet  18m                   deployment-controller  Scaled down replica set nginx-deployment-69c44dfb78 to 1
Normal  ScalingReplicaSet  18m                   deployment-controller  Scaled down replica set nginx-deployment-69c44dfb78 to 0
Normal  ScalingReplicaSet  2m13s (x25 over 29m)  deployment-controller  (combined from similar events): Scaled down replica set nginx-deployment-5d59d67564 to 4

更新 Deployment 的注意事项

清理策略

在默认情况下，revision 保留 10 个旧的 ReplicaSet，其余的将在后台进行垃圾回收，可以在 .spec.revisionHistoryLimit 设置保留 ReplicaSet 的个数。当设置为 0 时，不保留历史记录。

更新策略

.spec.strategy.type == Recreat，表示重建，先删掉旧的 Pod，再创建新的 Pod；
.spec.strategy.type == RollingUpdate，表示滚动更新，可以指定 maxUnavailable 和 maxSurge 来控制更新过程；
- .spec.strategy.rollingUpdate.maxUnavailable，指定在回滚更新时最大不可用的 Pod 数量，可选字段，默认为 25%，可以设置数字或百分比，如果 maxSurge 为 0，则该值不能位 0.
- .spec.strategy.rollingUpdate.maxSurge，可以超过期望值的最大 Pod 数，可选字段，默认为 25%，可以设置为数字或者百分比，如果 maxUnavailable 为 0，则该值不能为 0