Using Autoscaler to scale machines from 0 machine

The autoscaler project supports cluster-api. With this enhancement now the user can use cluster-api feature to scaling from 0 machine.

Settinng up the workload cluster

While creating a workload cluster, We need to set the below annotations to machinedeployment inorder to enable the autoscaling, This is one of the prerequisites for autoscaler.

apiVersion: cluster.x-k8s.io/v1beta1
kind: MachineDeployment
metadata:
  name: "${CLUSTER_NAME}-md-0"
  annotations:
    cluster.x-k8s.io/cluster-api-autoscaler-node-group-max-size: "5"
    cluster.x-k8s.io/cluster-api-autoscaler-node-group-min-size: "0"

Setting up the cluster-autoscaler

1. Clone the autoscaler repository

git clone https://github.com/kubernetes/autoscaler.git

2. Build the autoscaler binary

cd cluster-autoscaler 
go build .

3. Start the autoscaler

./cluster-autoscaler \
--cloud-provider=clusterapi \
--v=2 \
--namespace=default \
--max-nodes-total=30 \
--scale-down-delay-after-add=10s \
--scale-down-delay-after-delete=10s \
--scale-down-delay-after-failure=10s \
--scale-down-unneeded-time=5m \
--max-node-provision-time=30m \
--balance-similar-node-groups \
--expander=random \
--kubeconfig=<workload_cluster_kubeconfig> \
--cloud-config=<management_cluster_kubeconfig>

Note:

1. autoscaler can be run in different ways the possible ways are described here.
2. autoscaler supports various command line flags and more details about it can be found here.

Use case of cluster-autoscaler

1. Create a workload cluster with 0 worker machines
2. Create a sample workload

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: busybox
  name: busybox-deployment
  namespace: default
spec:
  replicas: 1
  selector:
    matchLabels:
      app: busybox
  template:
    metadata:
      labels:
        app: busybox
    spec:
      containers:
        - command:
            - sh
            - -c
            - echo Container 1 is Running ; sleep 3600
          image: busybox
          imagePullPolicy: IfNotPresent
          name: busybox
          resources:
            requests:
              cpu: "0.2"
              memory: 3G

3. Scale the deployment to create addition pods

kubectl scale --replicas=2 deployment/busybox-deployment 

4. Obeserve the status of new pods

kubectl get pods                                        
NAME                                  READY   STATUS    RESTARTS   AGE
busybox-deployment-7c87788568-qhqdb   1/1     Running   0          48s
busybox-deployment-7c87788568-t26bb   0/1     Pending   0          5s

5. On the management cluster verify that the new machine creation is being triggered by autoscaler

NAME                                        CLUSTER               NODENAME                                  PROVIDERID                                                                                      PHASE          AGE     VERSION
karthik-ibm-powervs-control-plane-smvf7     karthik-ibm-powervs   karthik-ibm-powervs-control-plane-pgwmz   ibmpowervs://osa/osa21/3229a-af54-4212-bf60-6202b6fd0a07/809cd0f2-7502-4112-bf44-84d178020d8a   Running        82m     v1.24.2
karthik-ibm-powervs-md-0-6b4d67ccf4-npdbm   karthik-ibm-powervs   karthik-ibm-powervs-md-0-qch8f            ibmpowervs://osa/osa21/3229a-af54-4212-bf60-6202b6fd0a07/50f841e5-f58c-4569-894d-b40ba0d2696e   Running        76m     v1.24.2
karthik-ibm-powervs-md-0-6b4d67ccf4-v7xv9   karthik-ibm-powervs                                                                                                                                             Provisioning   3m19s   v1.24.2

6. After sometime verify that the new node being added to the cluster and pod is in running state

kubectl get nodes
NAME                                      STATUS   ROLES           AGE   VERSION
karthik-ibm-powervs-control-plane-pgwmz   Ready    control-plane   92m   v1.24.2
karthik-ibm-powervs-md-0-n8c6d            Ready    <none>          42s   v1.24.2
karthik-ibm-powervs-md-0-qch8f            Ready    <none>          85m   v1.24.2

kubectl get pods
NAME                                  READY   STATUS    RESTARTS   AGE
busybox-deployment-7c87788568-qhqdb   1/1     Running   0          19m
busybox-deployment-7c87788568-t26bb   1/1     Running   0          18m

7. Delete the deployment to observe the scale down of nodes by autoscaler

kubectl delete deployment/busybox-deployment

kubectl get nodes
NAME                                      STATUS   ROLES           AGE    VERSION
karthik-ibm-powervs-control-plane-pgwmz   Ready    control-plane   105m   v1.24.2
karthik-ibm-powervs-md-0-qch8f            Ready    <none>          98m    v1.24.2