Red Hat OpenShift The Level Up Hour | Red Hat Livestreaming, 10 Nov 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: The Level Up Hour (E51) | Getting Started with OpenShift Single Node Cluster

Description

Let's deploy a single node Red Hat OpenShift cluster. In OpenShift 4.9, support for a single node cluster was added and in this episode, Bradd Weidenbenner, Red Hat Advanced Cluster Management Product Manager, joins to help us look at this new deployment pattern. We'll cover use cases and why you would need a single node OpenShift cluster, show you how to install it using the Assisted Installer from the OpenShift customer portal, and then walk through an overview of this new single node topology. What's the benefit of single node OpenShift? Single node OpenShift offers both control and worker node capabilities in a single server and provides users with a consistent experience across the sites where OpenShift is deployed, regardless of the size of the deployment.

Learn more: https://www.redhat.com/en/blog/meet-single-node-openshift-our-smallest-openshift-footprint-edge-architectures
https://cloud.redhat.com/blog/deploy-openshift-at-the-edge-with-single-node-openshift

Follow us:
Randy Russell https://www.linkedin.com/in/randy-russell-0312243
Jaafar Chraibi https://twitter.com/ChraibiJc
Scott McBrien https://twitter.com/stabby_mc

What is the Level Up Hour?
Every Wednesday at 9am ET we stream live demos on how Red Hat Enterprise Linux (RHEL) users and admins can benefit their organizations and improve their careers by learning how to use containers, Kubernetes, and Red Hat OpenShift.

Learn more at:
https://red.ht/leveluphour
http://red.ht/3ayHlHI

How to participate in the The Level Up Hour:
Watch on YouTube every Wednesday at 9amET
Chat with us on Discord https://discord.gg/5VMVGJt
Check out our GitHub repo https://github.com/level-up-hour/

About OpenShift:
Red Hat OpenShift is an open source container application platform based on the Kubernetes container orchestrator for enterprise application development and deployment.
https://www.openshift.com/
https://github.com/openshift/

#RedHat #OpenShift #tluh

A

Good morning good afternoon good evening and welcome to another working on openshift.

A

B

And good morning, good evening, good afternoon, whatever the case might be welcome to the level up hour, where we consider all things, containers and kubernetes and, of course, that wonderful, amalgam of them all the red, what way of doing those things openshift so like share, subscribe, announce it to the world so that people know that we're here and they'll want to know that we're here, because we have a really really interesting topic. Today, I am joined uh by my co-host jafar chiraibi.

B

um uh How are you doing today, jafar.

A

I'm doing much better than yesterday, actually so.

B

Well, we are relieved to hear that uh we were thinking that you might have to be doing this from bed with an iv drip, but uh but you're looking good sounding good. So uh today's show we might also be joined by brad wyden-benner, who is the red hat advanced cluster manager, product manager. We've had a few technical difficulties this morning, we'll see if, if we're able to have brad join us, but in the meantime we're going to be covering something that I think a lot of people will be interested in.

B

That's uh the open shift, single node cluster or what's sometimes called sno, although I'm very reluctant to introduce yet.

A

B

Three-Letter acronym to our world and so uh jafar. Maybe if you could tell us a little bit about about what is the single node open shift and then we'll get to why somebody might care about it? So what is it.

A

Yeah sure, so um um uh let me try my screen and we have a single slide that talks about it. So if you've been uh using openshift uh for some time, you know that with openshift 4 we had a topology where you needed at least three control planes. So that's, of course, part of the kubernetes, I would say architecture and at least two worker nodes. So the minimum was a five nodes cluster, but we had some customers who wanted to have a more compact installation of openshift.

A

So we worked on what we called the three node cluster, which came out, I believe with openshift 4 7.

A

So basically the masters and workers were co-located and it allowed you to have a a smaller footprint for some type of use cases.

A

But as we started um playing also into the uh telco 5g edge uh area, we we also had some customers asking for a very minimal footprint and basically they wanted to have a full-fledged openshift running on one single uh node and that's basically, the single node openshift, also known as sno.

A

So this is available starting from openshift49, and basically this is so there's there's been a lot of engineering to overcome those uh some some challenges because, uh as you might know, with openshift 4, we had a bootstrap machine that was controlling basically the flow of installing everything on uh the other.

A

On the other nodes, and basically we were using, uh I would say, containers to deploy, containers and doing this on. The same machine was not something uh easy to to handle, but thankful to our wonderful engineers, they've been able to work around that and have a single node openshift that not only does the bootstrap as it was done before, but once everything is loaded, it reboots and then it becomes a master, slash worker node.

A

So um it's um it builds itself, basically so yeah exactly yeah. So that's a a self installing uh machine, so we will be talking about the installation process uh uh just a bit later, uh but basically what uh the use cases for for this are mainly uh so. This is for production use. uh First of all, so we've had some uh other distributions, like code, ready containers which were aimed for developers basically so not for production use and those were not really a full-fledged open shift.

A

Now this is a full open shift running in a single uh node, so what you don't get with that is, of course, high availability, because everything is running on a single machine but other than that than that all the other features of a traditional openshift clusters are available even upgrades. You can uh go ahead and upgrade through the regular channels. So that's a great thing um that we we are. We are able to support that. um It.

B

A

B

Well, yeah, so, first of all, that's that's really sort of an interesting thing in that it's going to upgrade in a typical way and it will behave in some of the sort of management aspects like it's open shift right. So.

A

You're not going.

B

To have to do a different process, necessarily in terms of where the upgrades are managing your upgrades or anything like that.

A

Exactly yeah yeah.

B

Okay, we had a couple of questions already. I love this. We've already got questions in the chat, so satchel asked you know, and I know we're going to get into the particulars of install, but I figure we might as well answer a couple of these is.

A

Yeah, first of.

B

All you know is this uh something that can be installed on a vm, or is it only bare metal.

A

So uh so there are two uh two two sides: this uh answer: can this be installed on the vm? Yes, and that's basically, what I'm going to be doing uh but for support this is uh this thing this is for bare metal usage only so as an end user. As a customer, you will be using this on bare metal all right.

B

Well, and that kind of goes to some of the use cases that we're going to talk about here. um Mana canton had a a question as well saying: isn't um a single node openshift snow uh very similar to crc, which runs inside.

A

B

What's the what's the difference between the two.

A

Yeah, so um so, uh first of all, first one and that's a major one: sno is for production use, crc is for developers, so you cannot run applications workloads on crc, that's the the first thing. Crc is really a different distribution of of, I would say the openshift experience, but for a developer to have that printing on his laptop.

A

So I I mentioned the upgrades, because if you, if you've been using crc, you know that there's no upgrade.

A

You had to basically reinstall a new crc instance to get the the newer versions so um yeah.

B

Well, it's a developer piece and there's no need for it to be. You know continuously available, so production readiness really and again crc stands for code, ready containers. We have so many acronyms, you know so, um let's, uh let's talk a little bit about. You know more specifically about why this came into being, and I think uh you know, I think the thing that maybe makes the most sense to understand why this is important is to look at edge computing right, where you start to think about.

B

Okay, we're not talking about running things in a data center. Now we're talking about running something somewhere else and it's running by itself, and maybe maybe availability is a different question now, because the thing that's available can go away.

A

It might go away forever or it might be.

B

Something that can go away periodically and it's just not that big a deal so talk a little bit about how this fits into edge computing. First of all,.

A

Yeah, so um so, um if we speak about uh things like 5g telco edge, basically we're trying to run those uh virtualized uh networking function uh closer to the uh uh some not from a central send point but closer to uh to the end and customers, and um that's basically the the main um use case that that this is intended for so you're able to deploy a full, fully fledged openshift cluster on a single node uh and the uh so. Basically, some some customers want to have a portable, uh I would say environment.

A

They want their applications to be to to be to be available on a minimal footprint and, as you said, they don't want to have a whole data center just to be able to to deploy this uh small sets of applications that may or may not be uh business critical. uh So, for instance, you said they. These applications may may be working, not necessarily 24 hours a day, but maybe while you're in a restaurant or you are in a shop and you have some local uh applications. That are your point.

B

A

Yeah your point of sale, while you're up in in business hours, but not necessarily afterwards, so there are many use cases where things like that can can be suitable for this type of of uh deployments and basically, when you speak about edge, uh that's that's why we we made that type of uh new um new topology.

B

Right right well,.

A

So there's one other phrase.

B

That that has been sometimes used here uh that I think I'd like an explanation of, and that is frankensteining your own car yeah minimal costs. Okay, what on earth does that mean.

A

What on earth is that yeah? So um so I've been a uh an opportune solution, architect for many many years so uh for about five years and of course, uh for our own learning process and also sometimes, for you know, preparing pocs, preparing demos and stuff like that. We needed to have uh openshift clusters and uh going back to openshift3.x, we had what we called the all-in-one install and basically you had a full-fledged openshift cluster running into a single machine.

A

It could be a virtual machine, so since we moved to openshift4 it was something that was not available anymore, because now, if I wanted to have uh that openshift cluster running on my laptop now, I needed to have five vms running and, of course, the resources.

A

If you have 16 gigs of ram or or yeah even 32 gigs of ram, it's not gonna fit in there. So um so uh yeah. When I speak about frankenstein in your own car, say say it's: it means uh deploying uh a an snow to experiment some with some things to try new features to try. uh You know upgrades and things like that at a minimal cost. So, basically, you are not deploying uh five or six vms. You are just needing one machine and then you can try uh things like that.

A

uh Also, uh we speak about multi-cluster management. We have our solutions like acm, with the red hat, advanced cluster management solution. That allows you to manage multiple openshift clusters and say you want to do a poc or prepare a demo, and you want to manage many clusters say you want to manage three clusters.

A

So now the minimal footprint for your demo becomes 15 machines, whereas if we are using the sno, it becomes three machines right.

B

A

Yeah, it makes us much easier for us to um to to fool around with these things, to to learn uh also uh new things, without necessarily having a heavy infrastructure.

B

So you can say it's it, so what you're describing is sort of intermediate between code ready containers, which is very much a developer thing, not something that you can really use necessarily as a poc.

B

It is something that helps you kind of work on your app and see how it's going to work in a container, so this is sort of in between code, ready containers and then the full blown. You know highly available open shift and, like you said it's a difference between having five. You know having five things versus three things and all the resource requirements for having that smaller footprint right, yeah.

A

Yeah correct correct, so I I see well, I see we have some interesting questions.

B

A

Yeah, so um so so someone was asking about the minimum resources for this. So uh yes, of course, uh when we talk about when we speak about edge, uh it can be. You know many things if you are speaking about iot, uh it's a certain type of uh devices and it's a very, very minimum uh footprint.

A

But here we are speaking about uh servers that are going to be running uh business applications. So um so the minimal footprint for the sno is 32 gigs of ram and eight cores.

A

So that's that's, I would say the recommended minimum and um probably depending on your workloads, you might need more uh more than that so um yeah. So there are some questions about um the the installation. The requirements. Do we need dhcp, reservation and stuff like that. So uh these are some questions that we are going to tackle. When we speak about the installation so randy, should we go there or or should we.

B

I think we're oh, oh, I think we are. I think we are ready.

A

Let's, uh let's, let's, let's.

B

Leap into the live demo, what all right possibly go wrong.

A

Yeah exactly yeah, so um so first thing uh the purex are pretty uh straightforward, so you have, uh as with any openshift cluster, you have some um requirements regarding the api, so the api you have the wildcard for the apps for your cluster and uh so in a traditional openshift environment.

A

These are held by some bips, but for the sno. Basically, you need to have a dhcp reservation or a static ip for your virtual for your node uh and then those uh dns records are going to point to that uh to that ip.

A

So um you you need that an api dot cluster name that points to the sno static ip and you need the white card that dot apps. That points also to to the uh sorry to that uh ip address to that static ip address. So these are, I would say, the mean the main requirements.

A

um So in order to make it easier for you to install the single node openshift, we actually came up with what we call the assisted installer that supports the single node openshift installation and let me go ahead and share that with you.

B

A

Okay, so can you guys see my screen now.

B

Indeed, we can.

A

Okay, perfect, so what we're going to do? We are now at uh console.tradehard.com, which is our hybrid console uh for uh managing your openshift clusters for managing your subscriptions, etc. And, as you can see here, it says, create cluster. So uh it's something that we made to make it easier for you to create your openshift clusters.

A

And basically here we are going to create a cluster name. We are going to specify, which is the base domain, that we are going to use, and here we are going to mention that we want to install a single, node openshift, then afterwards, what this is going to do it's it's going to generate an iso for us, and this iso needs to be loaded on the machine that is going to be hosting uh the uh the the sno. So you can boot it via usb drive or pxe boot booting.

A

So, whatever suits suits, you will work so, for the sake of the demonstration, I'm going to be using a virtual machine, uh but keep in mind that this is not the production use case. So questions.

B

Asked earlier yeah.

A

Exactly we're going to start with an exception, exactly yeah, so just uh for the sake of time, I had already prepared a an iso that I have loaded on my uh virtual environment, uh and this is the the one here. So, as you can see in the ui, it says waiting for host and what I'm gonna do now is I'm going to create a machine on which I'm going to load the iso and we'll see how the installation goes uh goes on.

A

All right, so, let's go ahead, create the vm. Let's call it sno, and here I'm going to choose the local installation and choose the iso that has been generated.

A

So there's a minimal uh disk of uh 120 gigs and for the memory here we're gonna go and put 64 gigs of ram, although again uh with 32 gigs, you should be just fine. So I'm gonna change the cpu here and put it to the minimum.

A

And now I can start the installation process so, as you can see, it's going to boot, the core os uh instance and it's going to boot into the uh the bootstrap mode. So basically, it's going to to to load all the services that are used to configure the openshift cluster and at some point, when everything is loaded in the vm, we should have uh the the the machine showing up here.

A

um So basically it talks back to the to the to the ui, and this is very you know, easy easy for you to to get your hands on. So I just did an installation of a single node cluster this morning. uh It takes about 45 minutes to to one hour, to have everything loaded. There are multiple reboots uh needed because, as we said, since everything is running on the same machine, it changes roles, so it starts as a bootstrap. Then it reloads.

A

So now we see it entered into the discovering state and it's going to so it discovered that it's the machine that I had uh so it it asks for some ntp sources. So, let's.

A

Let's get give some servers over there.

B

I love that you just googled for an ntp source.

A

Yeah and work: that's it and.

A

So with this, what does it say? So if there are some uh configurations that are missing the the installer is going to complain about it. So, for instance, when you boot the machine it starts with the localhost hostname, so we're going to change the hostname there and we're going to call it sno and afterwards is going to get into a ready installation state and it's going to start deploying the components over there.

A

So, okay, let's go ahead hit next so now I I need to to select the network on which the installation is going to be.

A

Deployed and let's go ahead and hit the install cluster, so that's that's it! That's all I need to do and if we come back in 30 minutes in 45 minutes, maybe uh or maybe a little more or go on to have a fully running openshift cluster, so that's it. It's very easy to configure and uh and install, and once this is done, uh you have access to your cluster information.

A

Of course, while the installation is running, you have access to the logs and you can see the the state of the of the cluster once everything is finished. You get this nice ui that you know provides you the cube config to access the the console.

A

You can launch the openshift console and it tells you also what dns records you need to put in in there and so from here. We can access this single node openshift cluster and, as you will see, it's basically a fully fledged openshift cluster, except that, if I go to the nodes section, I see that I only have one machine and it pla. It plays the role of the control plane and the worker as well, and I have all the uh traditional features I can upgrade.

A

If I wanted to update to a different version and of course you can do everything that you so you have the monitoring all the dashboards are are available in there.

A

So it's really a fully working open shift environment uh except that it it's running.

B

Highly available.

A

Exactly it's not highly available. That's it right.

B

A

B

Some more questions here um so and we might be able to answer some of them so, first of all in terms of cpu, uh would it run on an older xeon or is eight cores all that's required.

A

uh So I I don't think that it's gonna look for some specific uh cpu set of instructions. So, yes, it can run on the xeon and actually I'm running it on the xeon and yeah. So again I don't know what the uh performance is going to be, uh but yeah it can. It can run.

B

And it might be that the application might not, you know, need performance, um how about uh you know sort of the subscription basis? uh This is basically a an openshift subscription, or is it a separate kind of subscription.

A

Yeah, it's it's! A traditional openshift subscription exactly.

B

Okay, um another question is that it, uh you know it seems like a small disc, uh you know, is it better if you use something bigger or you know,.

A

So um I don't know what you mean by a small disk. uh You mean the the 120 gigs of ram.

B

A

Yeah, so so you can, of course you can. You can put more, you can put more if you want to add installation. So it's it's, not a limitation. You can, if you wanted to go with 500 gigs for the disk go ahead and do that, but keep in mind that you can add disks afterwards, of course, you're not necessarily going to be resizing the disc, but.

A

I don't see why you could not uh actually to be honest uh but yeah. You can have this disc for the open shift. I would say components for the images for the registry and these things and if you needed more storage, you could ask some disks to to that machine and configure them and maybe even use the local storage operator to access those. uh Those disks.

B

Okay, so we also had the question that uh is this available: is this only open shift or is it something that's an okd.

A

um That's a good question, and actually I don't know if.

A

I I believe it's going to be on okd also uh I I don't have the answer for that actually, but I think you can do that with okay richo, I haven't tried it, but I I guess so.

B

All right, so um this is actually really instructive, and it's great to have all these questions. Keep them coming. um If you have them, uh was there something else that we should maybe talk about here. Where you know uh sort of coming up on to the top of the hour were a little short because we had a few technical difficulties but uh yeah. Maybe we should talk a little bit about uh the relationship of um single node openshift, to advanced cluster management and and some of the things that maybe are worth mentioning there.

A

Yeah sure so, um uh as we mentioned, the single node cluster is uh is intended for massive deployments. For example, you're going to be managing tens or even hundreds of clusters, and you you don't want to be managing those clusters by hand connecting to each console and upgrading the things deploying your applications to those clusters.

A

So this is why we have a very tight integration between acm and sno for many use cases. So, of course, sino is a regular openshift cluster, so it can be managed with acm as any other cluster. You can deploy your applications using the getups approach. You can deploy your policies and stuff like that. But what's interesting is that we can also install single node clusters uh using acm.

A

So basically, the discovery process that you saw is something that is available from acm. You can generate those iso files and you can have those single node, openshift clusters deployed uh from acm, uh so we we wanted to have the product manager uh here to walk us through some of those uh features. But unfortunately, uh as you said, we had some technical difficulties, but uh it's probably something that we can uh try.

B

For a future show.

A

Exactly yeah, we can have a more focused uh episode on on on acm and those types of use cases, but yeah, basically, acm can help you manage a fleet of single node clusters and upgrade them, deploy your policies, etc.

B

Yeah so um boating mcsquarepants, and I I want to read this question just because I want to say that name, uh you know, was looking for a little more clarity on this question about uh about dhcp versus static ip.

B

A

Because I think, when.

B

We start talking about that edge use case. That can actually be a pretty important consideration right.

A

Okay um yeah, so I believe you can specify a static ip in the boot configuration uh when booting the the core os instance. So you you have to enter into the.

A

Advanced kernel parameters and then you can specify your static ip, that you're going to use. If you don't have a dhcp server, you can choose to go with that with that option,.

B

Okay, well, so uh I don't think we have any more questions, but I want to go ahead and leave the floor open here. If anybody has some additional things that we haven't asked. uh Maybe if I have missed one, let's post, that there we go yeah, so somebody wants to try it out on aws.

B

Is there a doc or write-up available.

A

uh Yeah sorry, so there was a question about a acm. Yes, it's available in acm 2.4 uh for the.

B

A

Integration. uh Sorry: what was the other question.

B

Trying it out on aws.

A

So um well, basically, yes, there are some blogs that are available. We can share the links to those. Let me see if I can find the links.

B

Well, I imagine a lot of people would actually like to try it out through that that mechanism, um so any other questions.

A

Yeah so randy, I'm gonna post the link in the private chat uh and if you could restream, that would be good.

B

Okay, there we go coming up.

A

So, uh okay, there it is yeah yeah yeah, so I think yeah yeah, you already already shared it. So um it's not going to be a specific uh to aws ec2.

A

So this is a very, very generic installation process. uh Once you are able to load the iso into any any machine, so whether so, whether it's a virtual machine or a bare metal host, uh it's gonna be working.

A

The only thing to keep in mind is for production use. It's only supported on bare metal.

B

Right, I think we already covered the question about the xeon um yep and uh we've covered the question regarding the uh the licensing. This is a an openshift subscription and it really just comes down to a choice about how you want to deploy open shift yep all right. uh Let's see, um oh, you know actually yeah. I think we do you just covered.

B

The acm version is that correct.

A

B

All right other questions were concerns.

B

This is good. We had a lot of them.

B

Obviously this is a very hot topic, um and so you know apologies to everybody for a bit of a late start today um you know we uh we had hoped, as uh we've mentioned, that uh uh brad wyden benner, uh who is the product manager for acm, would join us, but you know what that's the opportunity for us to uh bring brad another day and uh really dig into uh acm, in particular, some of the newer features that are in there and we can even revisit uh sort of single node.

B

So again, uh if you find well one last question here from makoto, uh are there any plans to support horn.

A

A

I guess so that's fact I don't know.

B

So I mean that seems like I I don't know it seems like a logical thing to do given some of the particular use cases, but um you know I I think we might have to come back to that question in another day.

A

B

Right well so just another reminder if you enjoy and find value in the level of power. Yes,.

A

Yeah, so just to answer the arm question so openshift and arm, of course, uh is a thing we are working on on having uh openshift uh uh supported on on arm. I don't know where we are uh for that, but uh so for for the snow thing, I guess, as you said, that would be a logical step forward once we have full support for for the arm architecture.

B

All right, well so great session jafar. uh Thank you for walking us through this. This is a you know, a very compelling topic, obviously, and I think you know when we think about the impact of edge and how that's really becoming just more and more of a a priority out there in the world and we think about the coming tidal, wave, 5g and some of the implications there. I think uh you know the implications of uh single node openshift become very clear.

B

So with that, I think we we covered a lot of ground, we'll uh we'll come back to the subject of acm another day, but I just want to remind everybody. You know like subscribe and share the level up hour if you're, finding this useful join us for our next show, and uh thank you for joining us today. Thank you, jafar.

A

Okay, thank you very much randy and thanks. Everyone have a great day.

A