OVHcloud Managed Kubernetes service provides you Kubernetes clusters without the hassle of installing or operating them.
During the day-to-day life of your cluster, you may want to dynamically adjust the size of your cluster to accommodate to your workloads. The cluster autoscaler simplifies the task by scaling up or down your OVHcloud Managed Kubernetes cluster to meet the demand of your workloads.
Before you begin
This tutorial assumes that you already have a working OVHcloud Managed Kubernetes cluster, and some basic knowledge of how to operate it. If you want to know more on those topics, please look at the OVHcloud Managed Kubernetes Service Quickstart.
The easiest way to enable the autoscaler is using the Kubernetes API, for example using kubectl.
As explained in the How nodes and node pools work guide, in your OVHcloud Managed Kubernetes cluster, nodes are grouped in node pools (groups of nodes sharing the same configuration).
Autoscale is configured on a node pool basis, i.e. you don't enable autoscaling on a full cluster, you enable it for one or more of your node pools.
You can activate the autoscaler on several node pools, each of which can have a different type of instance as well as different min and max nodes number limits.
Info
In order to avoid unexpected expenses, you should be careful to not enable autoscaling on monthly-billed node pools. However, you are still allowed to do so if you know what you are doing.
A common configuration is to use non-autoscaled, monthly-billed node pools as base for your static workload, and autoscaled, hourly-billed node pools with smaller flavors for your dynamic workload.
Info
The source code of the following example is available in the public Github repository public-cloud-examples
When you create your cluster, you can bootstrap a default node pool in it, and you can add others in the Public Cloud section of the or directly using the Kubernetes API.
Deploying a test workload
Let's assume that you have created an MKS cluster with a node pool with its minimum set to 1 and its maximum set to 3.
In order to test the autoscaler, we offer you to install a Python heavy CPU load Deployment that deploys several instances of Python CPU load pods. The Python CPU load pod's goal is to consume all the CPU allocated to it. It's a CPU intensive operation but it uses a minimal amount of memory.
Create a cpu-load.yaml manifest for the python-cpu-load deployment:
As you can see, we will begin by deploying 3 replicas of the pod. Each replica consumes 150m CPU (0.150 CPUs), and we are using D2-4 instances, with 2000m CPU (2 CPU cores). In the tutorial we will increase the number of replicas to 12 then to 24, to see how the autoscaler grows up the node pool to 2 then to 3 nodes.
As you can see then with kubectl get nodepools, the autoscaler detects capacity has been reached and asks for a new node:
$ kubectl get nodepoolsNAME FLAVOR AUTOSCALED MONTHLYBILLED ANTIAFFINITY DESIRED CURRENT UP-TO-DATE AVAILABLE MIN MAX nodepool-d2-4 d2-4 true false false 2 1 1 1 0 10
And in a few moments, the new node is created and active, and all the pods are running:
The autoscaler will detect that the nodes are under the value scale-down-utilization-threshold parameter (the node utilization level, defined as sum of requested resources divided by capacity, below which a node can be considered for scale down, by default 0.5), and marks the nodes 2 and 3 as unneeded.
After some minutes according to the value of scale-down-unneeded-time (parameter that sets how long a node should be unneeded before it is eligible for scale down, 10 minutes by default), the node will be deleted and the cluster will be scaled down.
After 10 minutes we are back to 2 nodes:
$ kubectl get nodepoolsNAME FLAVOR AUTOSCALED MONTHLYBILLED ANTIAFFINITY DESIRED CURRENT UP-TO-DATE AVAILABLE MIN MAXnodepool-d24 d2-4 true false false 2 2 2 2 1 10
And 10 minutes later, we have only one node:
$ kubectl get nodepoolsNAME FLAVOR AUTOSCALED MONTHLYBILLED ANTIAFFINITY DESIRED CURRENT UP-TO-DATE AVAILABLE MIN MAXnodepool-d24 d2-4 true false false 1 1 1 1 1 10
Cleaning up
To clean up your cluster, simply delete your python-cpu-load deployment:
In this tutorial we have seen how to enable the autoscaler on a node pool on your OVHcloud Managed Kubernetes cluster, and how to use an example workload to test how it works.
Otherwise to skip it and learn more about using your Kubernetes cluster the practical way, we invite you to look at our tutorials.
If you need training or technical assistance to implement our solutions, contact your sales representative or click on this link to get a quote and ask our Professional Services experts for assisting you on your specific use case of your project.
