karmada-descheduler

descheduler规则

 `karmada-descheduler` 定期检测所有部署，通常是每2分钟一次，并确定目标调度集群中无法调度的副本数量。它通过调用  `karmada-scheduler-estimator` 来完成这个过程。如果发现无法调度的副本，它将通过减少  `spec.clusters`  的配置来将它们驱逐出去。这个操作会触发  `karmada-scheduler`  基于当前情况进行“规模调度”。
需要注意的是，只有在将副本调度策略设置为动态划分时，这个驱逐和重新调度的过程才会生效。这种策略允许根据可用资源将副本动态分配到不同的集群中。通过定期检查集群的状态和资源可用性，Karmada可以调整副本的放置位置，以确保资源利用的高效性和系统稳定性。

前提条件

确保所有k8s成员加入到karmada当中且karmada-descheduler加入到karmada-host中如图所示

测试

模拟缺乏资源而导致调度失败，创建俩个nginx副本将他们放到俩个成员集群中

```
root@karmada-cluster01:/home/karmada-teaching/descheduler# cat pp-deploy-nginx-descheduler.yaml 
apiVersion: policy.karmada.io/v1alpha1
kind: PropagationPolicy
metadata:
  name: nginx-descheduler-propagation
spec:
  resourceSelectors:
    - apiVersion: apps/v1
      kind: Deployment
      name: nginx-descheduler
  placement:
    clusterAffinity:
      clusterNames:
        - cluster1
        - cluster2
    replicaScheduling:
      replicaDivisionPreference: Weighted
      replicaSchedulingType: Divided
      weightPreference:
        dynamicWeight: AvailableReplicas
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-descheduler
  labels:
    app: nginx
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - image: 10.250.2.118:18000/library/nginx:latest
        name: nginx
```

kubectl apply -f pp-deploy-nginx-descheduler.yaml -n karmada --kubeconfig ~/kube-karmada
propagationpolicy.policy.karmada.io/nginx-descheduler-propagation configured
deployment.apps/nginx-descheduler configured

此时都在cluster01上 现在将cluster集群停止调度

root@karmada-cluster01:/home/karmada-teaching/descheduler# kubectl --kubeconfig ~/.kube/21config cordon k8s-cluster1-node2
node/k8s-cluster1-node2 cordoned
root@karmada-cluster01:/home/karmada-teaching/descheduler# kubectl --kubeconfig ~/.kube/21config cordon k8s-cluster1-node1
node/k8s-cluster1-node1 cordoned
root@karmada-cluster01:/home/karmada-teaching/descheduler# kubectl --kubeconfig ~/.kube/21config cordon k8s-cluster01 
node/k8s-cluster01 cordoned

删除nginx pod

root@karmada-cluster01:/home/karmada-teaching/descheduler# kubectl --kubeconfig ~/.kube/21config delete pod -l app=nginx -n karmada
pod "nginx-descheduler-7864466765-fgqsc" deleted
pod "nginx-descheduler-7864466765-lhvnr" deleted
pod "nginx-descheduler-7864466765-mbh4n" deleted

```

此时集群1的po状态

```
root@k8s-cluster1-master:/home/k8s# kubectl get po -n karmada  -owide 
NAME                                 READY   STATUS    RESTARTS   AGE   IP       NODE     NOMINATED NODE   READINESS GATES
nginx-descheduler-7864466765-b4h4l   0/1     Pending   0          18s   <none>   <none>   <none>           <none>
nginx-descheduler-7864466765-cvx4k   0/1     Pending   0          18s   <none>   <none>   <none>           <none>
nginx-descheduler-7864466765-xr22g   0/1     Pending   0          18s   <none>   <none>   <none>           <none>
```

大概等5-7分钟就会从cluster1到cluster2当当中

已经调度到另外集群当中