KubeVirt Tech Demos, 8 Feb 2018

Previous Meeting

⏯

youtube image

►

From YouTube: K8s Data Import and PVC Cloning

Description

An end to end demonstration using cinder/ceph. This demo imports data into a persistent volume claim and then clones that data into a second claim.

Automation Scripts: https://github.com/screeley44/setupvm-dev/
Data Importer Repo (Provided by auto scripts): https://github.com/kubevirt/containerized-data-importer

A

This is a multi-part demo on the walks through the automated deployment of seven Center on to new cluster. Following that we have seven all-in-one kubernetes cluster and demo, the Qbert data importer and cinder cloning. Provisioner automation here is gonna, be handled by custom scripts. We will provide things so scripts in the description. The first node we're gonna set up here. Is the sender know if we're gonna do this using scripts from the setup VM dev repository before we get started we're going to edit the setup. Config I've done this ahead of time.

A

The two options here that you don't want to set, or the Center hosts and the Ceph host I'm using AWS, so I'm setting these to the internal host names. After that, you're gonna want to mod X the setup CMU script and kick it off.

A

Now this scripts gonna take about 15 to 20 minutes, to run it's perfect time to go, get a cup of tea or coffee. Now, in the interest of time here we're going to accelerate the video and in just a second, the script will finish and there you go. We now have a OpenStack cinder and kubernetes node next, we'll be deploying Seth on our second AWS node. So now we're going to be setting up the stuff now and it's going to look almost identical to configuring, the OpenStack standard, kubernetes mode that we did previously.

A

The same configuration file needs to be edited here, meaning that the same repo that we use for the first one we need to also have clone down here.

A

Values. To note here, the sender hosts and SEF hosts need to be identical to the other node. They need to have the, in this case the AWS private host name for each of those nodes, we're also going to configure the setup type to see and VCF and the previous set up. This was a default value.

A

From there it's the same Jumonville sects, setup cmp dev and kick it off now. This group takes considerably less time to run and we should be done here momentarily.

A

Okay, so it's at this point that we now have a two node SEF cluster and we're gonna hop back over to our cinder OpenStack kubernetes cluster for the next step of the demo. So we've completed our setup of Ceph on our second node and now we're back on the OpenStack cinder kubernetes code. First thing we're going to do is deploy and all in one kubernetes cluster I'm going to do that. By going to the.

A

Kubernetes repository and deploying the cluster.

A

Now this is going to take over two five minutes, while this is going on we're gonna hop down to our second terminal. This is also shelled into the same machine. We're not gonna, be able to use the terminal that we're deploying the local cluster in and we're going to wait for that to come up.

A

So now, we've got a all-in-one grenades cluster running on next to our OpenStack Center deployment on our first AWS node for space constraints, I'm going to fullscreen this bottom pane. We do not need the top pane.

A

So, thanks to the automation script, we have a set of specs here to get our cinder, dynamic provision or deployed, and that's going to be handling the cloning operations for our images. There's some configuration that needs to go on and then with these I've done. In the background, if you're doing this yourself, please refer to the documentation, I may go ahead and spin up the provisioner here screen.

A

The we're going to deploy a secret, a storage class and native parishioner.

A

ah Sorry there we go and make sure it's running and make sure the storage class exists excellent. So now we've done that way. Is our data import service to fetch a file from a remote location and populate a PVC with it again. Thank you.

A

Thanks to the automation scripts, we've got a the repository cloned down in the root directory, so we're gonna hop in there and use the Matt fest provide for us again. There's some configuration that needs to go on in the background here. I've added a couple extra specs just for demonstration purposes. Those would be the checker pod and this tunic alone, PVC, aren't included in the repository. Those are just sanity checks for our demonstration right now. You can ignore the namespace spec. We are gonna, be doing this on.

A

They need to fall in space, so let's go ahead and create the imported pod, config and Porter secret and the importer, the importer pas PVC, and wait for this to start going. It's gonna take a second to the Benham, a provisioner to get that blowing them up the PVC. While we're doing that we're going to look at the config now the config file is where we're going to tell the data import service where we want our data that come from.

A

It takes two types of endpoints, either a straight URL if you are offering files straight through HTTPS or HTTP, or if using s3 protocols. We have any estimate client built into the damien ports RS. You can give the server endpoint and the bucket name and an object name that you want to bring it.

A

Let's check on our PVC, it's bound great, so we're going to worry the importer pod now right now we're bringing in just a zero byte file for time. We've tested this with as much as a couple two gigabytes and it works fine for us.

A

And it is pulling an image right now, we're just using an Alpine image and running by there and inside of that there we go and it will complete it now.

A

Next thing we want to do is kick off the cloning process, so what we are doing is specifying a PVC called the cinder clone. You can see in this case and the provisioner is going to be looking for PVCs with this annotation here sorry, that is hardly so. The annotation reads: que vas io / clone request and then the name of the PVC object. So our PVC in this case is called : PVC and it will be using these stimulants.

A

Your provisioner sorry small mistake here something important to note: the destination PVC, the the one with the clone needs to match exactly the storage capacity of the PVC being plumb.

A

Now we'll go ahead and rename this clone.

A

And create the now.

A

There we go and as a final check to make sure that the golden clone now contains a zero by file that we've imported it into the golden PVC. We're gonna execute this check upon the checker pod just contains a very long wait inside and announce the.

A

Golden, plum PVC.

A

And as soon as that is running, we can exact into it and poke into the directory. We go.

A

Oh sorry, because of the way we're doing this, we have to do it via the docker layer.

A

Now this isn't always the case. This is just a quirk of local, no cluster on AWS.

A

Ok, so now we are on the checker pond. We expect the data directory to contain our zero byte file and there it is hello Cooper. So what we've done is set up a two node AWS cluster we've deployed sender, OpenStack and kubernetes on one and Seth, on the other, we've plugged sender in to sell we set up a dynamic provisioner, that's been altered to handle cloning operations of PVCs and we've deployed that into kubernetes.

A

We created a PVC and populated that PVC with remote data using our data service and then using the clone annotation in our custom. Pvc we've cloned the golden PVC into our golden clone, and that is how we have managed to get this data populated from remote location in the kubernetes, and that is all thank you very much.