minikube // rebalance pods across the cluster farm

three nodes

    #minikube delete --all
    minikube start --driver=docker --nodes 3
    minikube addons enable metrics-server

fix the role labels for the workers

kubectl get nodes
    kubectl label nodes minikube-m02 kubernetes.io/role=worker
    kubectl label nodes minikube-m03 kubernetes.io/role=worker

three replicas

    #kubectl delete deploy web
    vi example-web.yml

apiVersion: apps/v1
kind: Deployment
metadata:
    name: web
    labels:
        app: web
spec:
    replicas: 3
    selector:
        matchLabels:
            app: web
    template:
        metadata:
            labels:
                app: web
        spec:
            containers:
            - name: web
              image: gcr.io/google-samples/hello-app:1.0
              ports:
              - containerPort: 8080
            affinity:
              nodeAffinity:
                requiredDuringSchedulingIgnoredDuringExecution:
                #preferredDuringSchedulingIgnoredDuringExecution:
                  nodeSelectorTerms:
                  - matchExpressions:
                    - key: kubernetes.io/hostname
                      operator: In
                      values:
                      - minikube
                      - minikube-m02
                      - minikube-m03

    kubectl apply -f example-web.yml
    kubectl get deploy -o wide
    kubectl get pods -o wide

see what happens when shutting down a node

    minikube node list
    minikube node delete minikube-m03
    kubectl get pods -o wide

==> the third replica goes to minikube-m02

restore the node and rebalance

    minikube node add
    minikube node list

    git clone https://github.com/kubernetes-sigs/descheduler.git
cd descheduler/
    kubectl create -f kubernetes/base/rbac.yaml
    kubectl create -f kubernetes/base/configmap.yaml

    #kubectl delete deploy descheduler -n kube-system
    kubectl create -f kubernetes/deployment/deployment.yaml

    kubectl get pods -o wide

==> back to normal

troubleshooting

    Warning  FailedScheduling  4m4s (x2 over 9m30s)  default-scheduler  0/3 nodes are available: 3 node(s) didn't match Pod's node affinity/selector. preemption: 0/3 nodes are available: 3 Preemption is not helpful for scheduling..

==> that was with default nodeSelector: role: worker

resources

minikube // rebalance pods across the cluster farm

three nodes

three replicas

troubleshooting

resources

multi-node

affinity

rebalance

moar / nodeSelector