Red Hat OpenShift Ask an OpenShift Admin | Red Hat Livestreaming, 21 Sep 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Ask an OpenShift Admin (E83) | Worker Node Bonanza!

Description

This week we're talking all about worker nodes - how to add them to single node OpenShift, how to create MachineSets with UPI clusters, and how to manually add nodes to IPI clusters!

Join Andrew Sullivan and Jonny Rickard Wednesdays at 11 a.m. ET for a full hour explicitly designed to help OpenShift admins. Each episode addresses your questions and covers the different tools that containers need to run in a cloud-native environment.

Want to learn more?

Try Red Hat OpenShift: https://red.ht/3uVxkya

Follow us:
Andrew Sullivan https://twitter.com/practicalandrew
Jonathan Rickard https://twitter.com/jrocktx1
OpenShift https://twitter.com/openshift

Subscribe to Red Hat's YouTube channel: https://www.youtube.com/redhat/?sub_confirmation=1

About Ask an OpenShift Admin:
Do you have questions? We want to answer them! Every Wednesday at 11am ET, hosts Andrew Sullivan and Jonathan Rickard sit down to answer your questions about Red Hat OpenShift.

About OpenShift:
Red Hat OpenShift is an open source container application platform based on the Kubernetes container orchestrator for enterprise application development and deployment.
OpenShift on redhat.com https://red.ht/3rYz5ZS
OpenShift on GitHub https://github.com/openshift/

A

A

Let's get into the big topic of Open Source.

A

B

Ecosystem really boomed.

A

C

Hello and welcome to this week's ask an openshift admin live stream. I am co-host. Andrew Sullivan joined, as always by my wonderful and amazing co-host. Yeah yeah we're co-hosts Johnny Ricard. How are you, sir I'm good man? How are you uh I can't complain, um uh actually I'm feeling a little proud of myself. I think this is the first of the streams that you and I have completely set up by ourselves. uh You know we we haven't had Stephanie.

C

In the background we haven't had Rohan in the background and I think I went in and staged everything um as near as I can tell people are people are watching us, so yeah, hello. Everyone uh well again welcome. Thank thanks for joining us, so yeah, it's um so a little bit of background there right. We we had for a long time.

C

We had a producer Stephanie and she was the magic behind this behind the scenes so like getting everything staged so that YouTube and twitch, and all that other stuff works and that our interface here is is set up correctly and all that other stuff and uh Johnny and I have done it a couple of times here and there by ourselves. But um we've always had you know adult supervision and yeah.

C

B

We don't have that yeah she's, always she's always done like the uh yeah, the the kicking of The Shins, underneath the table making sure that we do the yeah so the right things and do the right things at the right time. So.

C

Yeah yeah, you know hey, who would have thought that they'd, let us kids, you know, run run amok right yeah.

C

What's that, and what's that thing about, uh you know when the when the farmer's away the mice will play or whatever it is right, yeah, anyways, anyways, uh so I hope everybody's having a good week. So far, it's been a uh a busy week in the world of kubernetes in the world of openshifts uh Johnny and I. Like we've both been busy this morning there was a just a number of things going on. You had a customer meeting.

C

There was a company meeting like I've, been doing just all kinds of stuff, so it was only like 30 minutes ago right, you know just just before the stream started, we're both like. Oh no, what's going on with top of mind, and we came up with the whole list of stuff here. So there there's lots of stuff going on. In addition to that, today's topic is one that I think is pretty interesting um over the months and years.

C

We've gotten a lot of questions but, more importantly, there's been some confusion about how worker nodes work with you know: IPI and UPI deploy clusters. You know hey with in particular, I'll talk with a lot of account teams and customers, and they say: oh I want IPI I need IPI, I have to use IPI, but everything they're struggling because they want to deploy and configure IPI and a UPI like way so I usually ask like IPI is opinionated right.

C

It wants things a specific way, because that's how we make it easy, that's how IPI does its thing and they say. Well, we need the you know. We we need the ability to scale worker nodes. We need to be able to use machine sets. Well, you don't need IPI for that with on-prem deployments, really the only difference, the biggest difference between UPI and IPI is the integrated load balancer.

C

So one of the things that we'll be looking at today is how you can use machine sets with UPI to manage worker nodes, but we'll also look at the inverse, which is how to manually, deploy a worker node and join it to an IPI cluster as well, and then Johnny correct me. If I'm talking at a turn here but I, think we'll also be looking at adding worker nodes to single node, openshift, yep so and I I know, I've only got one one of those three things: you've got the other two. So thank you. Yeah.

C

um But yeah other than that. uh Welcome to another. You know, office hours live stream here on red hat live streaming, um so my you know our usual introduction here is. uh If you have any questions, if there's anything that comes to mind, anything you'd like to talk about, don't hesitate to ask us so whether you're on chat, if you're on social media right reach out to us, send us some sort of information or some sort of question. We'll do our best to answer those questions here on the Stream.

C

If we can't answer them during the stream, then we'll follow up either on next week's stream or you can reach out to us via the various method methods. Let me find the right thing to click here, so you can reach me at practical Andrew on social media, so whether it's Twitter, whether it's uh Reddit or whatever those happen to be I'm the same across all of those in addition to or in addition, we're also in the kubernetes slack.

C

So it's a great way to reach out and hit not just Johnny and I, but a whole bunch of other people and I made a post in that slack channel the openshift users slack Channel. Just before this started, you can also reach out via email. So this is also a great way to get a hold of us if you're not watching the stream live.

C

So if you're watching recording, if somebody has directed you to some portion of the stream to learn some information or something like that, don't hesitate to reach out to us with uh you know anytime. Whenever you happen to be watching it um so andrew.sullivan simply for me, first name dot. Last name: I know it's not that way for everybody, but uh for me it's andrew.sullivan redhat.com you can reach out via email, also throw Johnny under the bus.

C

uh So you can reach Johnny Jrock tx1 on most of the social media, I think you're uh Jrock TX, without the one on the kubernetes slack uh why at redhat.com or you can hit either one of us usually anytime. Somebody sends us an email, we always add the other one. uh So you know if, if you only remember one, uh that's fine too all right um so I noticed that my chat thing chat, yeah, see chatsby and weird again.

B

Yeah I can see it coming across on mine. Okay, um like I, just saw our hope9s reply and then it looks like it's catching everything: okay,.

C

So I'll keep an eye on it. It's um it's wonky, sometimes I've learned. So it's funny because you know behind the scenes so we're both now in the producer interface. So we can both like change things and click things and make it rearrange and all that other stuff. Historically, we've both been guests and the guest interface is much simpler and much less, uh not painful, crowded, distracting, like I, can see like multiple views of you and I can see multiple views of myself and all this other stuff.

C

So it's definitely you know my ad HD is oh yeah yeah all right. uh Let's talk about some top of Mind stuff. um Let me bring up our notes document here. What have we got going on?

C

um Let's see so a follow-up from actually I'm gonna share. My screen here uh um share.

A

C

Window I hope you're. The window I want yeah.

A

C

uh So a follow-up from last week, so last week we talked about um you, know scheduler and kind of ways that we can work with the scheduler manipulate the scheduler or rescheduling pods and stuff, like that and I talked about how I had a um a gist that I was using. That I was copying off of and all that other stuff and I said that I'd share that so I I have actually done that now, so we I got that published.

C

It's actually been published since last week, but I realized I, published it as a secret gist, so nobody could actually find it. So you know, uh but I will go ahead and post the link to this into our chat here. I'll use twitch just so that way you can distinguish it.

C

um So this is all the stuff that I used last week in order to kind of trigger evictions trigger out of memory terminations right, all of those types of things that were inside of that stream. So if anybody's curious, including you see, I, have links into the documentation as well as the uh sample pod, that I was using.

C

So you can see here. This is the uh the blog post that we talked about and the Pod that we used during that particular stream, so lots of good information in there. Hopefully, that's helpful to somebody we'll update the um YouTube and the other video descriptions to include a link to this, just as well as I just realized I, don't think I put a link to the YouTube video, no I didn't um up here at the top, so I will make at least one edit all right, yeah, you can see created seven days ago.

C

It's just been secret. The whole time yeah see there. The secret went away because it changed it all right.

C

um Let's see what was next on our list of things here, um so somebody on Reddit I think I replied to a thread where they were asking about how to learn openshift and how to get started and that type of stuff and one of the folks um kind of responded that openshift doesn't have developer entitlements.

C

That is simply put not not true, so I and in full disclosure. I didn't know this either until we had Tushar on the uh here on the stream back like this is 83. It was like 50 episodes ago. It was a long time ago uh and Tushar was was talking about. You know, ways that you can get and try openshift and all that and he mentioned developer entitlements, and that was I I news to me right.

C

uh So you can verify this if you'd like for yourself, if you go to uh so if we search for red hat appendix one and if we go here um we'll land here, but really we want this page oops foreign.

C

So if we look at this page, this is our license agreement page and all that other stuff. What we want is product dependencies and pick whatever your language is for appendix one. This is the one that we care about and if we scroll down here and we get down to table 1.2 development and production uses, so you can see here demonstration activities, individual coding and testing activities both fall under development use for all other Red Hat products, including exhibits, 1B, C and D.

C

So if we scroll down keep going here, that was exhibit 11a, we just flew by there and we get down to exhibits. 1B right, exhibit won't B includes openshift. Let me see if I can find the right table row here, but.

B

C

This is just sort of the yeah here we go openshift container platform, so this is just sort of the. The proof is a in the pudding. If you will around no really developer entitlements apply with openshift, you can use up to 16 entitlements to deploy a cluster, and you can use that for as long as you like to uh you know, there is no option in console.red hat.com to entitle it so it'll, stay evaluation, it'll just say expired, but that's perfectly fine, because you're using a developer entitlement and it will still get updates.

C

You can still install operators all that other stuff, just keep in mind that it is only for single user non-production use cases. So if Johnny and I are sharing a cluster, that's no longer single user right. If it magically accidentally turns into a production cluster, not that that ever happens right Johnny, that's no longer a uh developer cluster rate, it's no longer eligible for the developer entitlements, but so yeah that uh just keep that in mind again openshift, absolutely included inside of there. I also always recommend to folks. If you go to try.openshift.com.

C

It takes you to this landing page, so this will walk you through all the different trials and all the different things that are available, so I usually send folks to um to here. It doesn't include, or it doesn't highlight the developer entitlements. You have to go through the self-managed and then you can go through that way, but um it's a great place to get started and start uh start on the journey. So to speak, so managed services are especially fun.

C

You know, do the the Rosa CLI and it's like one command, assuming you have all the permissions right. It's one.

B

C

To stand up a cluster, it's great yeah.

B

It's a lot like I like how they did that too. It's a lot like setting up an eks or AKs. You know just a couple things and you're done.

C

All right, um let's see what other top of Mind stuff do. We have in here argocon, um what's happening with argocon yeah.

B

It's going on this week and um you know our our old teammate Christian he's out there uh representing uh you, know, he's doing. I was trying to find the link to his talk but um I'm sure it'll pop up on LinkedIn. So if I see it I'll put it in the in the YouTube comments, but yeah he's out there. So if you're watching online or if you're going to watch some of the scenes, make sure you check out Christian stuff, because uh you know he's always putting out a lot of good information about Kit.

C

Ops so and I guess: that's probably why he's not in chat right now, yeah.

B

Probably or maybe he'll, maybe he's gonna be late. You know, maybe he's got one hand on one hand off kind of thing. Yeah.

C

Yeah, okay, yeah conferences are not an excuse. I stream from a conference come.

B

C

Know Christian.

B

You set the bar so.

C

uh Let's see um there was one other thing: oh not really one other thing just a reminder of stuff: that's coming up. So, as is tradition right just a reminder, uh hopefully a gentle reminder to subscribe on whatever platform you're watching us on, because we do have a ton of content. That's coming up, uh so confirmed topics like things that we actually know are happening over the next three weeks. So next week, September 28th we'll be talking about hosted control planes, AKA hypershift with Adele zaluk.

C

uh So it's a really interesting one right using a cluster to host in containers the control planes for other clusters. So you can you know it's a resource, optimization thing: it's a speed to deployment thing right, there's lots of really cool stuff, that's happening! There. I think that one will be interesting, especially for any customers who have you know more than a handful of clusters.

C

um You know use ACM. You can deploy those quickly easily. You can apply policy based configuration I'm, not going to talk all through this we're going to be talking about it next week. uh Let's see the week after that, October 5th uh we'll be talking about pod Auto scaling changes that have happened in 4.11 4.10, in particular the alternative recommender.

C

So this was a new one to me um right so pod Auto scaling, Works off of a recommendation engine and you can create and use a different recommender to determine when it should actually trigger an auto scaling event. Now note: this is not the same thing as keita right so keita. If we go to uh kita.io.

C

Thank you, but IO is a random website. .Sh there we go um so keita is event, Driven Auto scaling. So this is one where, like it'll plug into I, think uh we have Tech preview support for it right now or Support Tech, so Tech preview isn't actually supported, but Tech preview integration where it'll hook into like Prometheus.

C

So you can say hey if this Civic metric for Prometheus is, is uh exceeds a value or is below a value. Do something so the alternative recommender, that's one of the things that we'll be covering is. Why is that different? How is that different from keita October 12th we'll be talking about composable openshift? This is another interesting one, especially for folks who are interested in um deploying a subset of the core openshift services.

C

So what I mean here is if you're familiar with the cluster version operator right anytime, you do an OC get Co I've got a cluster up here, OC git Co. So all of these things, let's say that maybe I don't want to deploy the marketplace right today. You don't have an option. You get the marketplace, whether you want it or not. Composable openshift means that we can make some modifications to these things inside of here. Some of these things inside of here so we'll be talking about that in three weeks time.

C

uh After that it gets a little bit fuzzier, but we do know that we won't be on the air the week of the last week of October I, think that's October 26th, uh because that's kubecon, so there'll be tons and tons and tons. I think I saw a list of sessions that are being done by red headers there and it's like 60.

C

B

There's it's massive yeah.

C

And you know it's kubecon: if you've never done or attended or or viewed virtually kubecon, um they aren't really vendor sessions. It's not hey! Here's! How red hat openshift does something better than everybody else. Rather it's all of our folks who are working Upstream talking about the stuff that's going on upstream and with all their various projects. So definitely you know we always take that opportunity, not just for Johnny and I to uh catch a breath, because you know we're used to doing this every week.

C

So it's not a big deal, but rather because there's so much good content. That's there that we, we just really encourage folks to go and check that out.

C

I think that was it I think that was everything um yeah top of Mind. There we go done. We did it. It almost feels like normal. uh Let's see anything in chat here.

B

uh Our hope9 is saying that we should get promoted.

C

I think that means to uh the uh Red Hat YouTube channel yeah yeah. That's a good point: Our Hope, nine. um We we try to update the calendar, um but honestly it's um it's on my list of things to do, and sometimes we forget so this morning, I sat down I, don't know you! You probably can't see that this is anything other than a yellow Square.

C

These are the like eight things that we do for each one of the uh each one of the streams, so you know create a post on Twitter create a post on LinkedIn, which I don't think we did today uh create a post on Reddit uh posting, the kubernetes slack post to the internal list right update the streaming like there's just all of these things that um as we're we're, alerting and we're getting better, um hopefully we'll uh be more consistent with those things. So Maya culpa, if you will yeah.

C

um All right, so, let's talk about today's topic.

C

um So, as I said at the beginning, you know, worker nodes are very thing right: everybody has worker nodes, everybody needs worker nodes um and I'm, going to I'm going to say worker nodes, but worker node doesn't just mean um a a node that is hosting your applications. It can also mean infra nodes right things that are hosting core infrastructure Services. It can also mean you know other types of of customized nodes. One thing that I think surprises some people is you, don't you're not limited to just three node types, right: control, plane, infra worker?

C

You can actually have as many different types as many different labels, as you like. So, for example, I can, have you know an infranode type that is specific, for you know, maybe in ingest or Ingress. Rather you know, I can have one that's specific for odf. If I'm using openshift data, Foundation I can have you know specific for metrics and logging right. So I you can divide those up. If you so choose, they don't have to all be one so long as they're, not hosting applications. They are considered infra.

C

Although I think the tools that we use on the back end do make it easier from a reporting standpoint to identify which ones need entitlements and which one don't if they, as long as they have an infrared tag or an infra label on the Node, but it doesn't have to just be infrared. It can be infra and worker and something else as well so yeah worker nodes, compute nodes, write nodes in general are something that we almost always deploy. You know and need, as a part of the cluster, so I'm going to start with UPI.

C

So if we have, if we look at our clusters here, so let's see get node so I deployed this cluster. This is a vsphere UPI. Cluster I deployed it as a compact cluster. To be clear. That is not something that is officially supported, so officially I believe that compact clusters, so a compact cluster, is one that is deployed with only three nodes where all three nodes are both control, plane and workers.

C

um Officially, it is only supported to deploy a new cluster with that model. If it is a non-integrated or a platform equals none cluster, it does, as you can see, work officially.

C

If you wanted to have a three node vsphere integrated cluster, you would want to deploy standards, cluster, so five nodes and then basically Mark the control plane as schedulable after deployment and then remove the worker nodes and then you'd end up with the same sort of config, but I skipped that process because I'm a rule breaker because I don't need I I, don't need my cluster to be supported so or my install process to be supported, so just anyways just be aware of that, and if we want to look um just for the sake of informational purposes, let's do uh my install config look like so you can see control, plane, replicas, 3, compute, replica zero and then, after that, I kind of I just did a standard deployment.

C

um So over here is my vcenter. You can see. I've got my three nodes running outside of here. Here's my bootstrap, which is currently powered off I, did the normal things to deploy the cluster. So if we look at.

C

Okay, a configuration script here, so all this is this is a very simple shell script that I use um so I just use govc in order to go in and configure each one of my nodes with the appropriate settings. So here attach the ignition config onto here, set the ignition config base or encoding rather set the static IP that I wanted to use for this and then do the normal, enable uid for each one of the nodes. So straightforward.

C

Here's all of my static IP information- that's up here, so this is all in inside the uh VPN inside a red hat stuff, all right, so we'll get out of that so standard cluster. um That's everything! That's uh that's! All I got running inside of here and like any um I guess. The last part of that is like any UPI. It is using an external load balancer, so I just have a h, a proxy deployed on my Bastion host and that acts as my uh my load balancer.

C

So it's UPI, like really trust me, it's UPI! So let's look at our cluster. So we can see here. This is our blue cluster, which is the one that we were just looking at. We can see that the infrastructure provider is vsphere, so this is really important when I deploy IPI the way that the installation process works is effectively openshift install. So the command line utility that we use will use the credentials that are in the install config.yaml. It will connect to vcenter.

C

It will upload a core OS template image and it will use that in order to create our control, plane, VMS and and our bootstrap VM. So it goes through and you know, bootstrap happens it hands over to the control plane.

C

The control plane takes over and then the control plane creates or I guess the installer creates a machine set, and then it scales that machine set to I think the default is three, but it would be whatever number of replicas are specified in your install config.yaml and that's how the initial set of worker nodes get created. So at the end of an IPI install I have a control plane with three nodes and I have a machine set that has been scaled to.

C

However, many worker nodes are specified and that's how the cluster gets deployed with UPI openshift install doesn't do any of that. Instead, we generate the ignition configs and then me, as the administrator I go in and as we saw in my script, there right I provisioned each one of these virtual machines, I configured it with the CPU and memory that I wanted I attached the ignition files using the standard methodology. So if we were to go here to look at the settings edit configuration so we can see here is our where'd.

C

It go anyway, somewhere inside of here. There's our static, ipconfig information. Here's our base64 I, don't know where it's at in here anyways. All of that information is attached, just like the installer would do with IPI I turn on my nodes, I, let them boot I, let the cluster deploy and, at the end of that I have and openshift a UPI cluster.

C

So how is this different? Let me find the right window here. So a couple of things so one if I look at my compute and go to nodes, we can see that I only have three nodes here. Just like we saw on the command line, my machine, so you'll notice, they're, all well, they're present they're all failed, and this is because that integration. You know that what the machine API does to provisioning control those doesn't apply with control plane nodes in particular.

C

However, recently openshift has begun, creating a machine set out of the box. I, don't know when this started, but historically that hasn't been there. So I was kind of surprised when I was getting ready for the stream to see that this. This guy was actually there and if we look at it, it actually has all of the configuration that we need right. You can see here, memory CPU disk, all of the stuff, so this may work right out of the box. I haven't tested it because again, I. Don't think that this is um I.

C

Don't think that this was there in previous versions, I think the last time I looked at this was 4.9 and it wasn't there so I I'm, assuming it was recent. So instead, what we're going to do is we're going to create our own machine set just a couple of other things here, as you can see, there's no Auto scaler configured health checks are there just like they always are, and my normal set of machine configs.

C

So let's create a machine set, so we'll get rid of the help over here, just to make it a little bit bigger and able to be seen so I've already created this I'm just going to copy it over from my um note document here and then we'll walk through what this file actually looks like.

C

So as with all things right, we start with all things. Kubernetes, we start with our our object, so I'm creating a machine sets I, give it a name. You can see, there's a bit of important information here that being what is our uh infrastructure name? Oh I guess we should start with the docs here we go so creating a machine set on vsphere.

C

Let me copy this and paste it into right here, so I'm going to paste a link to the docs in case anybody wants to follow along, um so it the docs here go over in an overview. Yeah yeah, um so sample yaml machine set for custom resource uh on vsphere, and it walks through all of these different options and everything that's available in here in the normal way that the docs does it also walks through all of the permissions that you'll need so remember. This set of permissions will need to match the same.

C

That is assigned to a a service account in vcenter that you would need for an IPI cluster Additionally. You still have all of the prerequisites for worker nodes for nodes that are provisioned for a machine set that match IPI. So, for example, DHCP still required right.

C

You can't do static, IP configuration with a machine set or a node, that's provisioned, via machine set, uh so that's important and why we wanted to talk about how to add manually provision nodes to an IPI cluster, because maybe you do have some things that you want to have a static IP. Well, that's how you would do it so in this instance, everything else is configured with a static IP, but these new nodes will rely on DHCP and then beyond that, it's just the normal stuff inside the docs.

C

So let's talk about our our object here, all right, so we give it a name. So we need to know this infrastructure name in theory, you already created that when you deployed the IPI cluster, so you saw here I have this in a folder that is named after that infrastructure name. If we look in the installation, Docs.

C

The sphere installing on these cluster, where is it installing a cluster on vsphere with user provision infrastructure?

C

C

So extracting the infrastructure name is the section of the install documents that we want here, and it tells us exactly how to do that and then, if we were to follow along down in the docs here, it tells us that hey, you need to create a folder inside of vsphere to hold your virtual machines so that storage, provisioning and all that other stuff works.

C

So we we should already have that information and we want to include it in the name of our machine, set, stop being overly helpful, those pop-ups that get in the way, um let's see here so yep machine API. We want to make sure it's there, blah blah blah. So if you're following along over here, all right, I'm just walking down through here right, you can see roll and infrastructure, ID and infrastructure, ID Dash roll, so I've just filled out all of that information here.

C

So you'll note here that I'm giving it a role quote: unquote, not of worker or infra, or anything like that I'm just calling it AOA again. That name is arbitrary down below so now, I'm in the spec, and just remember the way that this works is this is the machine set. This is the spec for the machine set right, so oops, again, overly helpful gets in the way and then once we get into the template is where this is.

C

What turns into the machine template that it that gets used so anything below here is what we'll see reflected in the machine's definition when it's created so down here when I create this metadata label. This is the label that will be applied to any nodes that are created not this. This is a metadata label, not a or a a metadatal label on the template, but not on the machine and that's hard to describe so this is the important one. If you wanted it to be an infra node right, we would just add.

C

If you wanted it to be, you know uh odf node right, we can add those as needed so down here, the provider spec. So one thing that I've found that's a little um I'll say irritating is there is no way to figure out what these are easily, so the provider spec is specific to each infrastructure that you're deploying the machine set onto, and it has you can see with.

C

You know, vsphere here things like the number of CPUs, the number of cores per socket, the amount of memory the disk assigned to it and all that so, while the docs and if we switch back over to the docs here right, the docs walk through and they tell you what some of these things are, but they don't tell us a lot about each one of those and what it means and as far as I can tell it's not in the docs at all, so instead I looked it up in GitHub, so I'll post this link into twitch as well.

C

So if we look at this particular page in GitHub, what we see is the definition, the API definition of that machine, vsphere machine provider. Spec, so switching back over here, here's our vsphere machine provider, spec object, type right, so I can walk down through here and again. I I get and it's a little bit frustrating you know. I I sort of understand go not well, um but you can walk through and see the comments in the code here to understand what each one of these is.

C

So sometimes people will ask things like: oh can I have more than one network adapter attached to my machine set provisioned virtual machines um with vsphere. Yes, so you can see here with network right. So it's a network spec object. If I scroll down here to find network spec, you can see it is an array of devices. So if I switch back over here and find my network devices, this is the first element in the array network name and then the name of that device.

C

If I had more than one networks that I wanted to attach it to I would simply put more than one of these in here.

C

It and it will iterate over those and it will add one for each one of those defined networks. Add a network adapter Peach, one of those Networks. On the other hand, if we look at things like oh I, want to add more than one disk drive to you know disk to each one of my nodes. If we look up here at disk gigabytes, you can see it's just an INT right. So there's only one value, that's available there. I cannot specify multiple disks for better for worse anyways.

C

You can. You can look through all of these docs or comments, I guess in the code, in order to figure out which what each one represents. If you have any questions, one thing that is kind of important the snapshot, depending on how Savvy and and how important it is to your vsphere team, maybe check out the difference between a full clone and the linked clones. They do have some implications on, particularly on the storage side.

C

I would say most of the time we want to leave it at the default of a full clone, but just check with your vsphere admins. If you have any questions there, so walking down through here, um one important thing that was also added is this snapshot, so I can specify a template, VM name as well as a specific snapshot on that template VM that I want to use if I so choose other than that everything else. So this workspace stuff all of this was inherited or pulled from the existing Integrations.

C

So, let's roll back a little bit so one when we deploy the cluster or so when we deploy the cluster it asks for with UPI. It still asks for all of the VMware questions right. It asks for what's the vcenter endpoint, what's the username? What's the password? What's the folder that you want to deploy to? What's the data store? What's the network right, it always asks those questions, whether you're doing UPI or IPI with UPI.

C

It's okay. If those don't work, the cluster will still deploy it'll still wait. You'll still have a fully functioning cluster at the end of it. Well, you'll have a mostly functioning cluster at the end of it, I should say what won't work is the Integrations, so it'll deploy it'll, say vsphere integrated all that other stuff, but what you'll have is um appear in the overview I'm going to open a new tab, so I don't have to lose any changes. I made there. What you'll have is alerts that are firing from the vsphere problem detector.

C

You know saying that hey the credentials aren't right or it can't do this or this isn't working or you know all of that type of stuff. So we want to be aware of we want to be. We need to make sure that the credentials have are correct. The configuration is correct and I can actually utilize the vsphere Integrations. If it doesn't work and again with UPI it will deploy. If it doesn't work, then provisioning machine sets won't work. If you need to change that, um I've got a bookmark for this.

C

um If it doesn't work, there is a KCs come on browser that describes how to go in and change those, so that KCs.

C

I, don't think I'm logged in no I'm, not so it won't actually show us, but that KCs walks through how to modify that configuration so go in and edit the vsphere.com it'll push out to all of the nodes. I think it will cause a node reboot for any nodes that are there so just FYI, because it has to lay the file out and then kublitz uses that file to do the integration. So it will trigger a reboot.

C

Sorry, I gotta I got a tickle I got this when I talk a lot. Apparently it causes me to cough.

C

uh Daniel Norris did you already address how you got dark mode um yeah not on this stream, but we did talk about in one of the other ones. uh It's pretty straightforward. Here you go to user preferences, so theme system default so I'm using a Mac. My my system theme is dark all the time, but you can change it right in here. If you so choose foreign, so pretty straightforward, but yeah 4.11 is where that was implemented.

C

um Yeah I see Johnny I think you said that.

B

C

All right so uh yeah. Let me re-click my thoughts here, um so you need to make sure that the vsphere credentials and everything are configured and ready and work again. Vsphere problem detector will alert you if that's not happening, uh and then the other thing is. If you were watching a moment ago, you saw it talked about how there's a template and we need to have this template available inside of our environment.

C

So I've already uploaded one in here, I'll just upload another one. Just to show you what it's like so I want to deploy an ovf template. So all I did here: I go to mirror.openshift.com.

C

You can also go through the uh regular customer portal access.redhat.com and go through the process of finding and downloading dependencies core OS for 11 latest.

C

Oh that's interesting latest took me to 4.11.0, but 4.11.2 is in here we'll use this one. So I just downloaded this OVA and I uploaded it because I don't trust the internet I uploaded it to a web server. That's running inside of the environment,.

C

Remember the IP address that I used here.

C

Here we go uh so yeah, oh see that one I clicked the same thing yesterday and I didn't notice. It apparently so we'll copy the link address for our OVA here I'm going to provide that into here.

C

We'll call this um blue dash.

C

With a different name, select compute resource, we'll use the cluster.

C

I think this step takes so long, because this is when it actually like pulls down more information to get better metadata about what needs to happen with the uh OVA deployment.

C

I've never understood why this step takes longer.

C

Is the okd UI like this? uh Yes, it's very similar um branding and stuff like that is going to be different. So instead of saying like red hat, openshift, it'll say Okay d, um but it it should be almost exactly the same where it might be for cluster. Go uh so template details, yep! We don't need to change anything here.

C

So the like, we don't need to set CPU. We don't need to set memory. All of that is determined by the machine spec.

C

What could be important here is the data store, as well as the virtual disk format, so you'll notice that there is no place to specify whether a one I want, whether or not I want my VM to have a thicker thin disk. It inherits that from the template that it uses here.

C

So, if I specify thin here, then any worker nodes that are created from this template will also have a thin provision. Disk Additionally, the the data score that I use here. Miter might not be important. It depends on your storage and it depends on how it does clones across data stores. So, for example, if I deploy this template into this osc3 VMware data store, but then over here I specify where's the data store uh data store.

C

So if I specify this AOS vsphere data store, what's going to happen, is it will clone across data stores with most storage in vsphere? There is no offloading of that right. It'll have to do a full like a copy of that actual template. Now, fortunately, the disk is only a gigabyte or something in size. But if you want to use offloaded clones, you know Vasa and all that other stuff make sure that it's in the same data store that you intend to provision the VMS from so Network.

C

This again doesn't matter because we're setting it in the in the spec.

C

We don't need to set these things because that'll be set by Machine API and we can click finish and then we should see is it'll go through and it'll deploy that so we're calling this blue our costs, so I will change that in my machine set and that's pretty much it so you can see credential secret. So this is standards. This is the same credential secret that'll get created during install time. We don't need to modify that number of CPUs. However many you want I want.

C

Two I only want two CPUs number of cores per socket. um I think the default is almost always one for this. It does have some pneuma implications, I think but use whatever works best for you or whatever the vsphere admins say amount of memory to assign to the VM the size of the disk to use for the VM. You can see I'm going with all minimums here, because my environment is resource constrained.

C

Whatever the network we want to use again, you can have more than one there the template to clone the virtual machine from because this is a machine set created with UPI that template won't be there by default. You will have to upload and create that template VM. It doesn't have to be marked as a template. It can be a regular VM or it can be a template, but you will have to upload that, whereas IPI does that automatically user data secret, this should be the same.

C

It's just inherited um from all of the others and then the workspace. You can see that all of this information comes from install config which gets put into the vsphere.conf, which is where it gets that data from. If it's not right there, then you just need to update it here or, and some of it you can change. So if you wanted to use a different data, store, um I think you can technically change like the data center and the folder and stuff like that. I don't think it's supported, though,.

C

um Do you need an on-premises cluster to set up okd? No, you can use um uh AWS. You can use Azure right, all of those just like with uh uh openshift. If you go to docs.okd dot IO, it has all of the instructions for how to deploy into the various hyperscalers.

C

All right, so, let's create our machine Set. uh Hopefully this guy's done by now yep create the machine set. That's it super simple. um So now, let's create a virtual machine, so real quick, you can see. I've only got still got the same three nodes that are over here, um bootstrap, which we're not going to use. This is one of our templates, the original one that I created yesterday when I was testing. This is the one that we just now uploaded.

C

So we'll increase our machine count to one click save, and now we can look at a couple of different things, so one I can go to machines. You can see that we have a new machine that is being provisioned, so it followed the naming template that we provided inside of the machine set. So you can see it has AOA in the name.

C

So we see our machine set here, has zero of One machines.

C

If I flip over to vsphere, you can see that I now have a brand new virtual machine that was created.

C

I, wonder if it shows in here.

C

I, don't know if it shows where it was cloned from.

C

Yeah there we go so here it says: blue dash, Archos clone two blah blah blah blah blah, so we can basically prove through the event log here that it did use this VM as a template and now we're just waiting for it to boot and do its thing. So it'll go through the process yeah. So it's booting up, it's getting all the stuff that it needs. So it'll go through the process just like with IPI and eventually we'll have a node.

C

That is it reports as provision as uh not just provisioned but joined to the cluster and it'll show up in the nodes over here. It won't show up, yet it won't. That's the last last step of it joining the cluster and doing all that stuff. So can you use this to move a whole cluster to a new VMware data store?

C

Yes, so that is one of the things that we one of the ways that that you can do that so effectively create a new machine set with a new data, store provision, new nodes into that new data store and then remove nodes from the original one.

C

um Where that gets iffy is with control, plane notes, uh so control plane nodes today cannot be managed via machine set. So instead you would have to do effectively a uh basically a fail right. You would manually fail and remove a control, plane, node provision, a new virtual machine in the new data store and then do a recovery of that of that control, plane node that was removed. We did a live stream on that a while ago. It's also in the docs. If you look in the docs.

C

Somewhere down here back up and res back up and restore control, plane and then replacing an unlvcd member. Basically, you follow this process, so that is one of the ways that you can do.

C

It um I think that is usually the default suggestion, even though, if I'm being honest that way sucks if it's an option to shut down the cluster, that might work as well so I say might because where it gets weird, if you will is if you're using PVCs that are provisioned using the VMware entry or CSI provisioner, those don't move so and if they do move, it breaks them. So this is why, if you do like a storage migration, uh it's not a live migration. It's a offline storage migration between data stores.

C

You want to make sure that the node is shut down so that no PVCs are mounted because normally a storage view motion would take all of the attached dicks with it to the new data store. We don't want that um so yeah you will have to if you do, need to decommission a data store and that data store is being used for PVCs you'll have to use something like the migration toolkit for containers and work with the app teams to move that data across.

C

Just be aware of that.

C

um But yeah other than that it would work absolutely for what you're suggesting there um and in fact especially I, would suggest if you're, just changing the storage for uh the worker nodes. So I would definitely suggest that um I.

B

C

It at this point we're just kind of sitting waiting for this thing to provision and do its thing and it'll eventually show up and join the cluster and be available to host workload, just like with any other IPI machine set type of thing.

C

um Yeah I feel, like I, went way long in that Johnny. Sorry, no.

B

It's all good, it's all good um yeah, so the the next two things that we're going to talk about like one is going to be adding a a node into a cluster, and um so I thought that this would be a lot easier on AWS. It actually is, and so uh yeah it's not gonna go well, but.

C

um You always say that, but by the way, I'll interject real quickly to point out that uh you know how is making those changes on the fly? Look, it showed up with all the roles that I added in there when it actually joined. Oh.

B

C

Yeah, it's nice when a plan comes together right, that's right!.

B

Yeah and and Maddie uh I think I'm saying your name right, hopefully I, hope, I'm, saying your name right. We called it Bonanza just because we we couldn't think of anything good and professional to say besides Bonanza, so that's that's kind of what we came up with no.

C

So I I did um you know we. We try to be cognizant of language differences and all that if I did actually do a quick search of what is the definition of Bonanza just to make sure that there isn't something that I was unaware of. So you can see a source of great wealth or Prosperity or a rich mind Vayner pocket of War. So it is a. um It is a valid word, not just a TV show from the 1970s here in the U.S um anyways, something that is very valuable.

C

I like to think that Johnny and I are very valuable right. Maybe a little bit of hubris in there.

B

That's right: that's right, mostly valuable.

C

Yeah not not entirely unvaluable how's that yeah many yeah, it's um anytime, you Google Bonanza, like if, without also definition, it's you'll find references to the TV show.

C

Anyway, sorry to interrupt Johnny, no.

B

You're good you're good, so um you know part of this is like we wanted to talk about. You know we have single node openshift and now we can actually add worker nodes to um a single node, openshift cluster. And so let me let me grab this link.

B

uh I have like a billion links open so.

A

I, don't know what that's.

B

Like I know, it's I I feel like I, want to say I'm better than this, but I'm not like it's just it's it's a mess. um Okay, so let me paste this um but yeah. So essentially, if you want to you, add your single node cluster, using the assistant, installer um and so just I guess. I could quickly walk through that. Let me go ahead and share my screen.

B

All right, awesome, oops and um yeah. So if you go, if you go into clusters, you can actually do a create cluster and then Data Center, and then you can just click on create cluster again and then you know just give it like a you know, a name. You know it's a demo cluster and then, whichever openshift version that you'd like and then you just select the install single node openshift and then you can do some static, Network configurations.

B

So if you, if you, you still need DHCP, but if you want to assign some static IPS uh to your node when it comes up, then you can do that. Just know that if you have two or more interfaces that are on the same VLAN and they're both coming up with the same IP it's going to uh when it's doing its checks, it's going to fail. So just keep that in mind, um but then yeah. No, it was doing it this morning. Okay,.

C

Yeah so um yeah, if you have two yeah two two adapters on the same subnet and yeah it'll it'll cause an error in the air is very non-specific. Yes, anyways.

B

Sorry, yeah no you're good. um So then, essentially what will happen is you'll, create that and then you get to a point where you can either download an ISO, for um you know like a virtual media. So like say, if you have a server, you can connect the virtual media and then do the remote install or you can download the full installation ISO that has all the stuff with it and um install your single node.

B

That way, and that's essentially, what I did so I installed that first single node instance with um the the iso and what happens is after a while it goes through the deployment, and then you get this nice interface within the assisted installer that has all of your stuff. And so now we want to add a host to the single node cluster. Now something to keep in mind with this is even though you're adding workers you're not actually adding High, availability or distribute.

B

You know like you're, not Distributing, the core components or the control plane across the worker nodes, you're, literally just adding worker nodes. So think of this, if you're used to kubernetes right, you have a single kubernetes control plane and then you add a worker node to it. You're, not adding any ha you're, just adding a worker, um so same concept and the the other thing with the the worker nodes is that they don't.

B

They don't have to have the same requirements from a uh compute resource perspective, right, I, think it's two vcpus or one physical, CPU and I I, believe it's like eight gigs of RAM and 100 gigs of storage. So it's significantly smaller than what you would need for a single node openshift, which is eight cores and 15 gigs by default, to use the assistant, installer uh interface and.

C

So snow still requires a non-integrated, install right.

B

uh Yes, yes, well, so you you can do some VMware integration. It's let me see if I can find it. They had the thing for VMware today.

A

C

I think it'll I think it'll offer you the option, but I don't know if it. If we support it or not, um eshwar yes, um you would need to have. You would need to entitle those additional worker nodes, um so I anyways, assuming it's a non-integrated, install you know with single node openshift. You can add those worker nodes on different infrastructures. So maybe my first, you know my single node openshift was deployed to you know a physical server.

C

You know, I can add a worker node, that's hosted in vsphere, or you know a hyper-v or whatever right, because hyper-v is a supported hypervisor for openshift.

C

um So you do have that option, at least in order to do that um it doesn't have to be on the same platform or the same infrastructure type yeah. So long as it's a non-integrated install.

B

Gotcha yep, and so so this was a it was just a bare metal, so I used my Dell server and um yeah. So we installed this now to add another machine to this. Cluster um I need to go back over here and then I'll just click add host. So if you see I'm in like the Clusters interface and then I just selected, my cluster that I created and then you just click on ADD host and then again you just download this uh um ISO and then you would mount it into a virtual media.

B

So what I'm going to do is I'm going to stop sharing this one.

C

So Medias um is single node openshift, a new offering from Red Hat.

B

It's not new, it's I, the the workers with single single node openshift is new I, think I, think 4.9. It went Tech preview or 4.8 tech review. Something like that.

C

Yeah it was 4.8 or 4.8 I think it was Tech or Dev preview. 4.9 was GA, maybe in a version earlier in there, but so single node openshift is like literally it's openshift like full open shift, just in a single a single node, as the name implies.

C

um So there are some modifications to reduce um like resources to be an appropriate size for a single node cluster, um so think things like it will turn down the resource requirements associated with Prometheus um or requests. I should say so that way more resources are available to Applications, because it doesn't need to account for a cluster. That could be sorry, there's a bug flying right in front of my face um it. You know it doesn't need to have enough CPU and memory for Prometheus to accommodate.

C

You know hundreds of nodes and tens of thousands of PODS that.

A

C

But yeah to Johnny's Point, um adding worker nodes is new in 4.11.

B

Yeah and so uh don't make fun of me because I'm using Windows here, but that's because I had to have Java and I didn't want to deal with Java on Linux, so I installed a Windows machine this morning. Why? Why.

C

Do I-lums always use like ancient versions of java dude.

B

And and Dell has done it like they've, essentially converted everything to Hydra or HTML5 and um you're like oh, okay, that's awesome, it just works so then yeah super micro, they're still using Java and it's a giant pain in the butt trying to get this to work on Linux. And so you know here we.

A

Are let me guess it's like Java, 5 or something you know, Java 8, so nice, nice.

C

Gross slightly slightly better than you know: Java 5, Java 6, but still ancient yep.

B

So yeah, so essentially you take that ISO and then you mount it into your virtual media, which is what I'm doing here um this just takes forever and then um what will eventually happen is we'll just boot from that ISO and then the machine will reboot and then it starts its Discovery and so we'll start seeing some of the um the metrics from the machine coming in.

B

So the number of CPUs we'll see the number of or the amount of ram, the amount of disks interfaces and stuff like that, um and then essentially just create, install cluster or install node, and um it won't work here because of the way that my home internet is so I can't actually modify DNS, so it can't actually find it. But the idea is the same. You just click create cluster. As long as it can talk to DNS, um then everything will be fine.

C

And uh many, um so it is heavier than code ready workspaces. Or what do we call that openshift local now openshift local yeah yeah, um because openshift local slash crw, does not include some features, functions: capabilities of openshift, because the expectation is it's just for developer, use cases and stuff like that. So they just like fully remove some things so I think code. Ready workspaces is four CPUs and 16 gigs of RAM yeah in the default config, whereas single node openshift is eight CPUs and 32 gigs of RAM yeah.

B

I missed the uh the boot menu so yeah. Here we are yeah. That's.

C

Okay but yeah, um so you went back through assisted installer and basically selected the add a note option and um had it generate an ISO, so it'll boot to the ISO. Does it pop into the assistant installer interface, or does it just automatically join the cluster.

B

So it'll pop into the interface like the normal machines do and then you have to click um install you know. So it's essentially just like doing the um the assistant installer but you're just adding it into the existing uh single node cluster.

C

Okay- and you have to do like I I guess- I should have mentioned this before just like with IPI and machine anytime. You use machine sets like when I was doing that with you with my UPI cluster. um Using machine set to add nodes, I didn't have to approve csrs. Do you have to approve csrs with single node openshift and assisted installer, adding a node? Yes,.

B

Okay, yeah, it's no middle and.

A

Apparently this.

B

Thing disconnects.

C

So JB Pratt you're right code, ready containers is the um the old name for openshift local code. Ready workspaces is the integrated IDE offering with openshift.

C

uh Many yep, you cannot I mean if you have a really large laptop, you could run it like resource High resource specs. For what it's worth it's. uh It is a issue or thing that we encounter here at Red Hat as well. Right. My like my laptop is uh uh I. Don't know it's an Intel, 8th, gen, I, think and.

B

C

Has 32 gigs of RAM, so it struggles to run code ready workspaces when I want to do something locally.

B

Yeah and the thing about the code or the open stuff local is that you can override the defaults, um but just know that you're going to take a performance set because everything's on that on that config. So um you know your control, plane, worker and everything is there, so it will take a hit.

C

Yeah uh Christopher: can you build storage classes? I assume you mean with code ready containers, um yes I, think so, um but it wouldn't be able to mount external storage unless the VM that it creates is connected directly to the outside world.

C

um I, don't know I I, don't know enough about code, ready containers or openshift local to know how that works, but it I mean in theory right. It's, it's a real openshift, just with some things removed, but um storage does work from what I remember it's just. If it's running on a bridged, Network behind Nat on your laptop then like you, wouldn't have a way of mounting that so far as I know. Maybe.

A

C

Wrong um yeah now I feel like I should go and test it.

B

All right, so this is going to take a little bit to check in but yeah. So essentially the the machine will start checking in and then, if we go back and we look at our console again uh so we'll see the node here, it'll pop up and it'll have all of its specs, and so then um it goes through a set of tests that it runs to make sure that the host is going to.

B

um It makes it sure that the host will, you know, is valid and um once that's done, there will be a little button over here and you just click install node and then it'll go through and it'll do its installation and then it'll be part of the single node cluster. So after you approve some csrs.

C

Yeah I I think the important part you know that we talked about. Is you, don't add worker nodes in the same way that you would with like a UPI or another cluster right? You wouldn't you wouldn't go and install core OS and you know point it at the ignition file and do all that other stuff. Rather you use assisted installer and have it generate the iso and add it that way. It's it's a much simpler process.

C

B

C

Many uh no Johnny uses Fedora I. Think right for your.

B

Main yeah I use Fedora mainly but I I do have Windows in this instance because of java, and it's just because it's if you've ever dealt with like Java dependencies and all that stuff I, just I I, don't want to deal with it. So it's easier in Windows and a VM that I can just burn.

C

Yeah uh Dimitri uh openshift was a great tectonic alternative. um Just kubernetes running I mean aren't they all uh I. Think k3s is more evolutional than openshift.

C

um I would say that they target different use cases, different perspectives, um yeah to JB Pratt's point they solve different problems, um so openshift is kubernetes, like literally when you deploy open shifts right openshifts um is is built upon kubernetes. What makes it openshift is red hat takes the kind of full stack of things, so the operating system is Rel core OS, which is a a flavor. If you will of red hat Enterprise Linux, then we Deploy kubernetes.

C

On top of that, and then we deploy operator lifecycle manager and install all of the things that make openshift into openshift, which is really like a set of add-ons Services capabilities right built on top of kubernetes. So think you know the admin console right. It's just a service, just a set of PODS, that's running an openshift. The ability to manage the nodes you know from within the cluster itself is machine, config operator right deployed on top of that kubernetes cluster itself.

C

So k3s is really really interesting for a number of reasons. Not the least of which is it's Ultra lightweight right. It's really good at managing. You know, with the bare minimum amount of stuff in place, the ability to deploy kubernetes based applications So within the red hat portfolio. We're working on a thing called micro shift um is that right, Johnny yeah.

A

C

Think they're working on a better name for it than microshift, but similar concept of being able to deploy. uh Basically, you get to choose the operating system that you deploy and then you Deploy microshift on top of it, and it provides that openshift kubernetes API on top with only the services that you need.

B

B

C

So so way back when um so what? Four years ago, five years ago, Red Hat bought a company called core OS core OS had a Linux distribution named container OS and a management interface for kubernetes called tectonic.

C

So when Red Hat brought and at the time red hat had openshift three, which was which used Rel six I, think at the time, maybe seven as the underlying operating system and.

C

um We had the openshift admin console and all that so oh and atomic limit Linux right. We also had Atomic Atomic clinics. uh So when we bought core OS, we basically took all of the really cool things that they were doing with tectonic and container Linux and merged them together and so tectonic became a part of the openshift console container, Linux and atomic basically merged together to create what is today Red Hat Enterprise Linux core OS.

C

uh So it is a Rel kernel in Rel packages, but it's packaged up and used using rpmos tree and with many of the management tools that came from container Linux and core OS.

C

So um Johnny I know we're running uh kind of long today, I feel like I've, been running my mouth a lot so.

B

um It's all good yeah.

C

So uh do you I, don't know how much time you've got left.

B

Well, yeah so I'm just waiting on this node to finish and then um yeah basically like. If you want, we can punt until like next week or we can do it as part of the blog for the the UPI part. But it's up to you because I I, just I, have to bounce like here in a nice few minutes. Okay,.

C

Yeah so um Let's, uh let's push off um because so the the last thing that you were going to show is how to do a how to manually provision a worker node and join it to an IPI cluster. So yep, um let's do a blog post. We can record that um I think I mentioned last week. Maybe we maybe we could do a short or something like that.

C

I've been talking with our marketing folks, you know I last week or was it the week before whenever time ball wins together, when I was with the marketing folks, we were talking about different different things that we can do so. Maybe we can do something like that if we can find time but um yeah and I know, we don't always have time to run super late, um let's see just reading through Comics or comments here rather um Dimitri, so I could already use infrastructure and it costs money.

C

um Yes, and no so remember everything, red hat does is open source, so the Upstream for openshift would be okd. You can absolutely use okd and it is, uh as I said earlier, like 99 the same right things like branding and all that's going to be this going to be different, so it won't say red hat, open shift in the corner, it'll say Okay D, but of course you don't get support So.

C

If you're familiar with how Centos operates now right, it is always moving forwards that type of stuff, oh Kitty, is the same, and we can have a lengthy conversation about that. Someday um yeah, if you, uh if you do, choose to use openshifts, just like with all red hat products, right, you're, you're, paying for supports and you're paying for all of the all of the things that we do, there's a great um if you search for uh Red Hat, the value of uh uh subscription like there's a way better than I could ever do.

C

You know set of things that explain exactly what red hat does and the value prop, and all that um I'm in The Bu I'm, not a salesperson.

C

uh Let's see red hat becomes a part of IBM Fedora, silver, blue user, nice, so RPM os3 um yep, that's if essentially or exactly what happens with uh core Os rel4 Os underneath the covers.

C

So if you were to look inside of an an openshift cluster or an okd cluster, the way that it does operating system updates, is it just pushes out a whole new RPM, OS tree layer and then activates it in the future, and one of the future shows that we've got plans is thing, is a thing called core OS layering, so you'll be able to customize your own core OS to add packages, drivers Etc so same exact principle, right fuel you'll, be able to deploy and update that uh those real core OS nodes using rpmo stream or or.

B

C

B

C

Excited to get those guys on I.

B

Did I share with.

C

You the the slide presentation that they shared with me yeah. They.

B

C

Really cool stuff I'll share that with you yeah.

B

That's awesome.

C

um Wireguard service reports- um I, don't know, that's a good um sorry wire guard I know we're talking about openshifts. Usually that makes me think of I. Think it's Calico offers the ability to do wire cards in between nodes. I. Don't remember one of the sdns offers that all right uh I know you got to go kids and all that other stuff, Johnny yep.

B

I just wanted to quickly show like when the when the node checks in to the console. This is what you see, but the reason why I have insufficient here and it's because it's just 18t U-verse doesn't allow me to delegate DNS or do anything funky with DNS. So it's a pain but anyways.

B

um If, if I had my DNS set up correctly, then this would just work and then I would just click install ready host and then it essentially approve some csrs and then it's all done.

C

Very cool yeah, it's um the the big change is just using assisted installer. uh You probably could get it to work in the traditional way, but why assistant installer makes it so much easier. Yeah.

B

Yeah I was gonna, I was gonna, cheat and like pull the boot ISO down and then do it that way. But then I was like.

C

Why make life hard on yourself Johnny.

B

Yeah yeah I know.

C

All right well, thank you. Johnny I appreciate your uh your efforts. Getting all that stuff set up. Sorry, we didn't get to the second part um and be able to show all the work that you put in there but, like I, said we'll make a recording of it, we'll figure out something, um even if we steal some time from next week.

C

uh So thank you to everyone who has joined us today. We really appreciate your time and attention. um Thank you for everybody. Who's been chatting. We love chat. We love to answer questions. That's the whole reason we're here. So uh thank you for the folks who are here live if you're, not watching us live, um don't hesitate to reach out um or if you want to follow up with us and and have conversations about anything that we're talking about here.

C

Offline um you're welcome to reach out, so you can reach me at practical Andrew on like Twitter on Reddit on all of the various places. You can also reach me via email, andrew.sullivan.com uh and, of course, I always encourage, encourage folks to go to the openshift users, um kubernetes Channel kubernetes team, yeah channel on the kubernetes team, yeah yeah. Well, I'm trying to think of slack terminology here, they're all different um Jenny. Do you have a link to that by chance.

B

Offhand but I will really quickly just one.

A

C

Copy copy link haha I got it, got it: okay, all right, I'll, post this into twitch, so it'll get propagated out. uh So there's the uh the link to that kubernetes channel. If you want to chat in there uh throwing Johnny under the bus, you can also reach out to Johnny as well at Jrock tx1, because he's in Texas um and or johnnyjohnny redhat.com. So thank you so much everyone. We really appreciate your time uh again. Don't hesitate to reach out at any time tons of stuff going on in the in the upcoming weeks.

C

So if you're not subscribed to the channel already definitely encourage you to do that. You can also go to red.ht. Slash, live stream. I'll put that into this chat.

C

So if you go to that link, that brings you to the landing page for Red. Hat live streaming, including a link to our streaming calendar, as our hope9 called us out on earlier. We'll do a better job about updating what the topics are, but you'll at least have reminders of when we're going, live and stuff like that.

C

So have a great week. Everyone uh thank you for catching up with us and Johnny I'll. Leave you with last words.

B

Hey man thanks for everything, uh it's an awesome week, looking forward to the next. You know few weeks uh I talked to Dean from VMware, so uh we'll get him on as well. So um yeah we got a lot of good things coming up so looking forward to it.

C

C

uh Oh Johnny I forgot what we're supposed to do. We got to click, I got it I got it. You know I'm sitting here staring at you.