Kubernetes不带to-revision的kubectl rollout undo存在无限死循环问题
Posted cnskylee
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了Kubernetes不带to-revision的kubectl rollout undo存在无限死循环问题相关的知识，希望对你有一定的参考价值。
第一步，编写首次部署所需的yaml文件，包括三块：命名空间定义、服务定义、部署定义，nginx的镜像版本为 1.16.1
内容如下：
apiVersion: v1
kind: Namespace
metadata:
name: load-balance
---
kind: Service
apiVersion: v1
metadata:
labels:
app: nginx-1161
name: nginx-1161
namespace: load-balance
spec:
type: NodePort
ports:
- port: 80
targetPort: 80
nodePort: 30012
selector:
app: nginx-1161
---
kind: Deployment
apiVersion: apps/v1
metadata:
labels:
app: nginx-1161
name: nginx-1161
namespace: load-balance
spec:
replicas: 3
selector:
matchLabels:
app: nginx-1161
template:
metadata:
labels:
app: nginx-1161
spec:
containers:
- name: nginx
image: nginx:1.16.1
imagePullPolicy: IfNotPresent
ports:
- containerPort: 80
protocol: TCP
# 执行创建
$ kubectl apply -f ./nginx-deployment.yaml
namespace/load-balance created
service/nginx-1161 created
deployment.apps/nginx-1161 created
# 第一次创建的容器实例我们简称为“7d”
$ kubectl get pods -n load-balance
NAME READY STATUS RESTARTS AGE
nginx-1161-7d57d5c7c4-prk5p 1/1 Running 0 22m
nginx-1161-7d57d5c7c4-s4pqk 1/1 Running 0 22m
nginx-1161-7d57d5c7c4-v42kt 1/1 Running 0 22m
# 测试访问Nginx的主页，成功返回Nginx欢迎页面的html代码
$ curl http://192.168.223.201:30012/
<!DOCTYPE html>
<html>
<head>
<title>Welcome to nginx!</title>
<style>
body {
width: 35em;
margin: 0 auto;
font-family: Tahoma, Verdana, Arial, sans-serif;
}
</style>
</head>
<body>
<h1>Welcome to nginx!</h1>
<p>If you see this page, the nginx web server is successfully installed and
working. Further configuration is required.</p>
<p>For online documentation and support please refer to
<a href="http://nginx.org/">nginx.org</a>.<br/>
Commercial support is available at
<a href="http://nginx.com/">nginx.com</a>.</p>
<p><em>Thank you for using nginx.</em></p>
</body>
</html>第二步，执行第一次滚动升级，修改Nginx镜像的版本为1.18.0，增加滚动升级的模块配置
apiVersion: v1
kind: Namespace
metadata:
name: load-balance
---
kind: Service
apiVersion: v1
metadata:
labels:
app: nginx-1161
name: nginx-1161
namespace: load-balance
spec:
type: NodePort
ports:
- port: 80
targetPort: 80
nodePort: 30012
selector:
app: nginx-1161
---
kind: Deployment
apiVersion: apps/v1
metadata:
labels:
app: nginx-1161
name: nginx-1161
namespace: load-balance
spec:
minReadySeconds: 5 # 增加配置
strategy: # 增加配置
type: RollingUpdate # 增加配置
rollingUpdate: # 增加配置
maxSurge: 1 # 增加配置
maxUnavailable: 1 # 增加配置
replicas: 3
selector:
matchLabels:
app: nginx-1161
template:
metadata:
labels:
app: nginx-1161
spec:
containers:
- name: nginx
image: nginx:1.18.0 # 配置修改
imagePullPolicy: IfNotPresent
ports:
- containerPort: 80
protocol: TCP
# 执行滚动升级
$ kubectl apply -f nginx-update-I-deployment.yaml
namespace/load-balance unchanged
service/nginx-1161 unchanged
deployment.apps/nginx-1161 configured
# 查看滚动升级的状态
$ kubectl rollout status deployment nginx-1161 -n load-balance
deployment "nginx-1161" successfully rolled out
# 查看副本集（rs，即ReplicaSet）
$ kubectl get rs -n load-balance
NAME DESIRED CURRENT READY AGE
nginx-1161-56dd96d66c 3 3 3 84s
nginx-1161-7d57d5c7c4 0 0 0 26m
# 查看pods和服务
$ kubectl get pods,svc -n load-balance
NAME READY STATUS RESTARTS AGE
pod/nginx-1161-56dd96d66c-724rg 1/1 Running 0 119s
pod/nginx-1161-56dd96d66c-cnkn6 1/1 Running 0 81s
pod/nginx-1161-56dd96d66c-zhq4c 1/1 Running 0 119s
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/nginx-1161 NodePort 10.96.92.254 <none> 80:30012/TCP 28m第三步，执行第二次滚动升级，修改image的版本为 1.20.1
apiVersion: v1
kind: Namespace
metadata:
name: load-balance
---
kind: Service
apiVersion: v1
metadata:
labels:
app: nginx-1161
name: nginx-1161
namespace: load-balance
spec:
type: NodePort
ports:
- port: 80
targetPort: 80
nodePort: 30012
selector:
app: nginx-1161
---
kind: Deployment
apiVersion: apps/v1
metadata:
labels:
app: nginx-1161
name: nginx-1161
namespace: load-balance
spec:
minReadySeconds: 5
strategy:
type: RollingUpdate
rollingUpdate:
maxSurge: 1
maxUnavailable: 1
replicas: 3
selector:
matchLabels:
app: nginx-1161
template:
metadata:
labels:
app: nginx-1161
spec:
containers:
- name: nginx
image: nginx:1.20.1 # 配置修改
imagePullPolicy: IfNotPresent
ports:
- containerPort: 80
protocol: TCP
$ kubectl apply -f nginx-update-II-deployment.yaml
namespace/load-balance unchanged
service/nginx-1161 unchanged
deployment.apps/nginx-1161 configured
# kubectl get rs -n load-balance
NAME DESIRED CURRENT READY AGE
nginx-1161-56dd96d66c 2 2 2 3m46s
nginx-1161-7cccdf449f 2 2 0 17s
nginx-1161-7d57d5c7c4 0 0 0 29m
$ kubectl get rs -n load-balance
NAME DESIRED CURRENT READY AGE
nginx-1161-56dd96d66c 2 2 2 4m2s
nginx-1161-7cccdf449f 2 2 1 33s
nginx-1161-7d57d5c7c4 0 0 0 29m
$ kubectl get rs -n load-balance
NAME DESIRED CURRENT READY AGE
nginx-1161-56dd96d66c 1 1 1 4m8s
nginx-1161-7cccdf449f 3 3 1 39s
nginx-1161-7d57d5c7c4 0 0 0 29m
$ kubectl get rs -n load-balance
NAME DESIRED CURRENT READY AGE
nginx-1161-56dd96d66c 1 1 1 4m13s
nginx-1161-7cccdf449f 3 3 2 44s
nginx-1161-7d57d5c7c4 0 0 0 29m
$ kubectl get rs -n load-balance
NAME DESIRED CURRENT READY AGE
nginx-1161-56dd96d66c 0 0 0 4m15s
nginx-1161-7cccdf449f 3 3 2 46s
nginx-1161-7d57d5c7c4 0 0 0 29m
# 至此，滚动升级全部完成（数据抓取时间延后，非升级速度慢）
$ kubectl get rs -n load-balance
NAME DESIRED CURRENT READY AGE
nginx-1161-56dd96d66c 0 0 0 26m
nginx-1161-7cccdf449f 3 3 3 23m
nginx-1161-7d57d5c7c4 0 0 0 52m检查历史版本的详细信息：
$ kubectl rollout history deployment nginx-1161 -n load-balance --revision=3
deployment.apps/nginx-1161 with revision #3
Pod Template:
Labels: app=nginx-1161
pod-template-hash=7cccdf449f
Containers:
nginx:
Image: nginx:1.20.1
Port: 80/TCP
Host Port: 0/TCP
Environment: <none>
Mounts: <none>
Volumes: <none>
$ kubectl rollout history deployment nginx-1161 -n load-balance --revision=2
deployment.apps/nginx-1161 with revision #2
Pod Template:
Labels: app=nginx-1161
pod-template-hash=56dd96d66c
Containers:
nginx:
Image: nginx:1.18.0
Port: 80/TCP
Host Port: 0/TCP
Environment: <none>
Mounts: <none>
Volumes: <none>
$ kubectl rollout history deployment nginx-1161 -n load-balance --revision=1
deployment.apps/nginx-1161 with revision #1
Pod Template:
Labels: app=nginx-1161
pod-template-hash=7d57d5c7c4
Containers:
nginx:
Image: nginx:1.16.1
Port: 80/TCP
Host Port: 0/TCP
Environment: <none>
Mounts: <none>
Volumes: <none>
测试回滚到上一个版本：
$ kubectl rollout undo deployment nginx-1161 -n load-balance --to-revision=2
deployment.apps/nginx-1161 rolled back
$ kubectl get rs -n load-balance
NAME DESIRED CURRENT READY AGE
nginx-1161-56dd96d66c 3 3 3 135m
nginx-1161-7cccdf449f 0 0 0 131m
nginx-1161-7d57d5c7c4 0 0 0 160m如果不指定--to-revision，则默认回滚到上一个版本
$ kubectl rollout undo deployment nginx-1161 -n load-balance
deployment.apps/nginx-1161 rolled back
$ kubectl get rs -n load-balance
NAME DESIRED CURRENT READY AGE
nginx-1161-56dd96d66c 0 0 0 138m
nginx-1161-7cccdf449f 3 3 3 135m
nginx-1161-7d57d5c7c4 0 0 0 164m个人理解的回滚机制：
通过测试发现，不带 --to-revision 参数的 undo 回滚，会进入死循环。如上图所示，会始终在V2和V3版本之间来回的切换，而无法回滚到最早的 nginx-1161-7d57d5c7c4 版本。不知道这算不算K8S回滚机制存在的一个Bug呢？
以上是关于Kubernetes不带to-revision的kubectl rollout undo存在无限死循环问题的主要内容，如果未能解决你的问题，请参考以下文章