Introduction to kubernetes resource objects and common commands (II)

Deployment

Deployment seeing the name of a thing one thinks of its function , Deployment means deployment .Deploymen Used to control the Pod, send Pod Have multiple copies , self-healing , Ability to expand, shrink, etc .

Multiple copies

copy , It can be understood as Pod The number of . these Pod Will be randomly assigned to any node in the cluster .

Create a named my-app Deployment of , Use image as nginx. Replications for 3.

• Command line mode

kubectl create deploy my-app --image=nginx --replicas=3

• yaml The way ,my-app.yml

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: my-app
  name: my-app
spec:
  replicas: 3
  selector:
    matchLabels:
      app: my-app
  template:
    metadata:
      labels:
        app: my-app
    spec:
      containers:
      - image: nginx
        name: nginx

kubectl apply -f my-app.yml

Expansion and contraction capacity

Expanding or shrinking capacity is to expand or reduce the original number of copies . When the traffic is large, the capacity can be expanded , After the flow peak, the volume can be reduced , Maximize resource utilization .

• Command line mode

kubectl scale --replicas=5 deployment/my-app

The number of copies is determined by 3 A into 5 individual .

• modify yaml The way

kubectl edit deployment my-app

take replicas Change it to the quantity you want , Save and exit to expand or shrink the volume .

self-healing & Fail over

Use Deployment The deployed application is always consistent with the set number of copies , When pod Exceptions or server downtime occur , Or delete pod, The replica will be pulled up again on the normal node , To meet the number of copies we set .

We simulate the k8s-worker1 The node is down .

Find out k8s-worker1 Two of the nodes pod Status as Terminating. stay k8s-worker2 Two more Running State of pod. Failover is achieved .

Expand ：

Look inside the cluster pod Information .

kubectl get pod -owide

Monitor the cluster pod situation , Print details .

kubectl get pod -w

threshold ：pod An abnormal condition has occurred ,k8s Will not kill immediately pod To restart , Instead, the corresponding restart operation is performed only after the threshold is exceeded , To prevent frequent startup caused by faults such as short-term network problems pod.

Scroll to update

The traditional updating and upgrading requires that the service be stopped for upgrading .k8s Don't stop the service , Will pod Replace... In turn Become the latest , That is, start a new pod, Kill an old pod, Until all updates are completed .

take nginx Image update to 1.16.1 edition .

• Command line mode

kubectl set image deployment/my-app nginx=nginx:1.16.1 --record

• modify yaml The way

kubectl edit deployment/my-app

take images It is amended as follows 1.16.1 in , Save to exit .

Expand ：

View a deployed yaml.

kubectl get deployment [my-app] -oyaml

Version rollback

Rollback the version to a certain time in history .

• View history

[[email protected] ~]# kubectl rollout history deployment/my-app
deployment.apps/my-app 
REVISION  CHANGE-CAUSE
1         <none>
2         kubectl set image deployment/my-app nginx=nginx:1.16.1 --record=true

[[email protected] ~]# 

• View a history detail

[[email protected] ~]# kubectl rollout history deployment/my-app --revision=2
deployment.apps/my-app with revision #2
Pod Template:
  Labels: app=my-app
 pod-template-hash=5ff664f457
  Annotations: kubernetes.io/change-cause: kubectl set image deployment/my-app nginx=nginx:1.16.1 --record=true
  Containers:
   nginx:
    Image: nginx:1.16.1
    Port: <none>
    Host Port: <none>
    Environment: <none>
    Mounts: <none>
  Volumes: <none>

[[email protected] ~]# 

• Roll back ( Back to the last time )

[[email protected] ~]# kubectl rollout undo deployment/my-app
deployment.apps/my-app rolled back
[[email protected] ~]#

• Roll back ( Back to the specified version )

[[email protected] ~]# kubectl rollout undo deployment/my-app --to-revision=2
deployment.apps/my-app rolled back
[[email protected] ~]# 

The rollback version should be filled in as required .

Summary ：

except Deployment,k8s also StatefulSet 、DaemonSet 、Job etc. Type resources . We all call it The workload . Stateful applications use StatefulSet Deploy , Stateless applications use Deployment Deploy https://kubernetes.io/zh-cn/docs/concepts/workloads/controllers/

This paper is written by mdnice Multi platform Publishing