Kubernetes Cluster Deployment with Nutanix Karbon

  • 29 March 2019
  • 0 replies
  • 4391 views

Userlevel 7
Badge +35

Are you interested in deploying production grade Kubernetes clusters in your on prem data center? Well, with Nutanix Karbon, you can quickly provision, manage, and operate your Kubernetes clusters all with the Nutanix Prism. Let's get started. Now that Karbon is general availability, we'll just gonna go ahead and click on the menu button, and then down to services, and then Karbon. If Karbon is not enabled in your environment, there'll be a button here to press to enable it and it'll get spun up in about 5 to 10 minutes as a couple of docker containers which lives on the Prism central VM.

Since mine is already enabled, I'm gonna click on the link to take me to the Karbon console. Now that that's opened up, I'm gonna go ahead and create a Kubernetes cluster.

We're gonna get started here working through the development cluster workflow. All of these settings are configurable, as we'll see. I'm gonna actually come back and deploy a production cluster, but just to show you the development cluster workflow. First, I'm gonna name the Kubernetes cluster. This is for the Prism Element cluster you wanna deploy this Kubernetes cluster onto. I only have one PE registered to this Prism central, so I only have the one choice. The Kubernetes version, I'm gonna leave the most recent. For the host operating system, I do have two versions.

I'm gonna leave the older version selected so I can show a host OS upgrade in a later demo video. Now, I'm gonna hit next. We can see here these worker resources are configurable, both from the number and the amount of memory or VCPU or the storage. Master and etcd are locked down for the development cluster. There's really only need for one, but you can change the resources.

As I mentioned earlier, I'm gonna go ahead and go back so we can see how the production cluster looks like. We see here the default is three workers and for the master the default is active/passive. You can also, if you want more than two VMs, you can put it behind an external load balancer. You can choose anywhere from two to five, although that does require you to enter in the external load balancer IP. I'm just gonna leave it active/passive for this demo. The number of etcd DVMs, you can choose three or five and obviously tweak the resources as well.

I'm just gonna leave the defaults for everything in this deployment. For the network, right now the only CNI we support is flannel. We'll be adding more in future releases. For the service and pod IP subnets, you can likely just leave these defaults. Essentially, the only requirement is that these do not overlap with your physical networking. So, unless your physical hosts are in the same subnet, you can just go ahead and leave those as default.

Finally, is the storage class. The default storage class that gets created is built off of the Nutanix CSI driver and Nutanix volumes. This is gonna enable us to easily provision storage to our Kubernetes workloads. We do have to enter in the Prism Element information. Again, if you have more than one Prism Element registered, it would show up in this dropdown. You can choose which storage container to use for this.

I'm just gonna leave it as the default container. Also, your reclaim policy. What do you want to happen to these volumes once the user is done with them? It's either delete or retain. The file system, I'm gonna leave it as a default ext4, but you can choose xfs.

Finally, I'm gonna leave enable flash mode untoggled, but generally if you have a couple different storage classes, you might wanna create one that does have flash mode enabled, depending on what sort of workload you're running. Finally, I'm gonna hit create. This is gonna go and start the Karbon Kubernetes cluster deployment. We'll see this percentage slowly increment. It should take about 5 to 10 minutes to deploy, so I'm gonna go ahead and fast forward the video now.

We now see the Karbon demo Kubernetes cluster was successfully deployed and is healthy. I'm gonna go ahead and download the cube config. Come over to my terminal. Move my cube config into my desired directory. I'm gonna go ahead and do cube control get nodes. We see my two master nodes. We might have three worker nodes. Now, this cluster is ready to go and I can deploy any Kubernetes workloads that I desire.

It's also CNCF compliant, so I can add on any of my favorite third party add-ons, some of which we do install for you, so stay tuned for an upcoming video where we'll go over that. I hope this was helpful. Thank you for watching. We're gonna have a couple more Nutanix Karbon videos coming soon.

👉 Are you running Nutanix Karbon? Be the first to hit reply and continue the conversation on this topic.

0 replies

Be the first to reply!

Reply