 Good morning. Good afternoon. Good evening wherever you're hailing from. Welcome to another edition of OpenShift Container Storage Office Hours. Today, we are talking with Michelle De Palma about auto scaling multi cloud gateway. That sounds super fun. I like scaling stuff. Okay, so I have an idea. It's just an idea because I haven't run through this already. So I'm going to propose that something and you can tell me if you think it's absolutely horrible. And if you don't want to do it. Sure. All right, so starting with, obviously, we have the latest and greatest OCP cluster going, which I assume you've got yours up already. Yeah, my clusters here somewhere. Yeah. Okay, so OCS latest and greatest install, which we can do again that's fast, very generic install we're not doing anything special with the install. It's also hit the operator will just go through that. And then someone from the, the MCG Nuba team because I was like, Oh my God, I got to do office hours. What should we do. So he said, he suggested why don't you do quay. And you see me pausing because up in Canada we call it key, but let's install the quay registry. It will use Nuba, it will use the MCGM point. And then, so, so what we can do, and we can verify that we can show how you can just go see it using creating buckets and stuff like that. But if we scale down the horizontal pod auto scalar to something really stupidly low. And then if you can help create a deployment that like spins up a thousand pods or something ridiculous with an always pull on the image we should be able to see the scalar in action. Okay. Just to let everybody know I'm having like the worst technical difficulty morning, and quite some time so this will be fun for the record. I like the idea. I like the idea. I don't know if we'll be able to pull it off so show folks probably will need your help today so yeah this is an office hours though so feel free to ask any questions related to storage. To be honest with you my my weakness here would be the registry part, like the details on, like, do you want to do let's encrypt and upload our own certificates would you do it insecure there's some things there I'm like, Oh, what is the easiest path with yeah like so that how do we spin that up quickly. Okay, that's an operator. Right, but you know when you go yeah can you share your screen like when you go to push and pull from your new private registry you have to tell open shift about it. I think you have to, if you're going to do it insecure you have to modify one of the cluster objects so to me that's the unknown. I'm like there's some shenanigans around that shenanigans. Okay, but if you just spin it up you can see it using the new band playing you can see it. Yeah. So you see the screen you're you see that I'm on the I'm on 4615 I don't think that'll make as big a difference. What update is available. Yeah 4616 shouldn't make that big of a deal. Let's see. Alright, so operator hub. Here we go. And folks if you have any questions or can't see anything or have any issues with what is on screen. Let us know I can maximize sizes and fix things. So, how would I install the operator registry, I would you wait you have a CS installed yet. No, I have a rotational disc problem. Yes, so I have to, we're going to have to jump through a lot of hoops to get this working I hope you know that. So, let's see, OCS. First, I need the local storage one. I have local storage and it registered rotational storage, no less, which super sucks which means I have to go in and patch it. We've done this on a show before but it's been so long. I'm not sure remember how to do it. So we'll see. Yeah, this is like a fresh cluster. I had to blow this away a couple weeks ago and rebuild it and it's kind of just been sitting here waiting to be abused. So yeah. Okay. Yeah, give me more reasons to do stuff with it. Yeah, I like that. So, do you use let's encrypt. I do. I am actually I have a let's encrypt hoodie hanging up in that closet over there. Yeah, you've good swag. Yeah, I have to say yeah. Well, today I'm just wearing my Nike Detroit shirt because you know Detroit haven't been there like an actual downtown hub to feel the energy of the city and a long time. So yeah, I'm kind of missing it. I don't recall this storage operator taking so long, but let's see. So what, okay, so I have the problem of rotational discs, right. Okay, so this local storage registry is going to pick up my SSDs. Because this is, this is, you know, seven nodes running on one machine. Okay, so seven VM running on one machine it picks up the SSDs as rotational. But maybe with this is a very new version of the operator so maybe it fixes. Let's see. I mean it just came out. I'm reading this correctly, the 30th of January. So yeah. It's doing stuff folks you can see it here. Let's, let's pop up on that tab real quick and see what's going on in here. See it's installing oh it just finished perfect okay. So, local volume discovery, I don't know if I need to create a discovery here local volume set what do I need to do with the local volume anything anything I don't think so right. Oh, God. So, first off, I haven't had my first couple I don't think so. It's been a while. Yeah, and it's early. So we'll try it without and see what happens I can't remember having to do anything but that was was. Let's see. All right so this is 462 so we got some updates. We definitely want to enable the operator recommended cluster monitoring on this name space. I'm running for six so stable for six. Yep, all this looks good. Check box folks you when you are installing OCS I highly recommend you check the check box because if you don't you won't have good metrics observability on your storage and that's like it should be selected by default. I will mention it some good point that is that is a very good point. Yeah, you should have that out of that. Yeah, actually I should mention that on a lease show later today that hey by the way, I know it's the operator writers that would have to correct it but it's it's one of those things where it's like, if we recommend it, it should be checked. All right. So we can see what's going on with this one here. Seven nodes on one bare metal on one seven and one. Okay, seven VMs, three of them are sized for OCS and they're already installing. Yeah, so I got three control nodes, you know, just whatever size three nodes size specifically for OCS and then an extra worker node, just cause. Okay, so this is so you know that it's okay so just so you know I was thinking of auto scaling because the when I was on for the first time before someone brought up performance and so I thought from from the customer point of view, which you need to be concerned with this, does it doesn't scale right like can we can we manage any upbursts and of activity so it just I just thought it'd be very meaningful to show auto scaling and what knobs you can turn to make that something for your environment that's also but I hope to have some more information about like sizing and stuff like that there's some anyway there's some articles coming out but they're I can't discuss them and to be fair I haven't read them yet so I can show you. All right, so CS appears to be installed. Okay. Storage cluster keys. Yes. All the defaults are just leave unless you want to go smaller, or yeah, whatever, whatever. I usually do too. Oh, let's just do half and then we can scale up if we need to work. Okay, so put it on these nodes. Because as you see, these are the nodes with eight CPUs. It says minimal cluster will be deployed the selected nodes do not match OCS storage cluster requirements of an aggregated 30 CPUs and 72 gigs of RAM I don't know how that's possible. Because they all have 64 years of RAM. Okay, just so no you're okay because you're just slightly like it's like 61 I think you should proceed. Okay. Yeah, this is a tech preview warning. That's weird. Anyways, alright, so I do have to do internal attach devices though don't I. They are. Yeah they are. Well you do all kinds of stuff I don't do an AWS. I didn't AWS right like that's why yeah like I'm like oh this is going to work. Let's try. So let's see you'll select our notes here. With worth it. What are we going to call this volume set just LV 0000 how about that. Cool. Alright, just type. Yeah, I'm just looking those advanced. I'm just looking at the initial here block. The OCS cluster requires a memo three nodes for the initial appointment only one node match the selected filters. Really. That's weird. Only two nodes match. Okay now it's all three cool. Okay, yeah, it's like checking it all out it was checking it on the background that's cool. Yeah, good for API is like API is working that's awesome. And as you see we switched from that internal screen on the, you know this toggle on the left that you would normally use right, or external if you're, you know, using some kind of S3 bucket of some sort. So since I'm super sophisticated and this cluster is in my basement sitting on top of a refrigerator. Yes, I'm very, very technical here. And if you've watched the show previously, you would have noticed that we have done this before and have had issues but now it works. So let's see. Next. As long as we're not doing anything that make Chris Blum cringe. I don't know he's in chat. So, we'll see a bucket will be provided. I guess I select this one. Okay. What's it, wait. What's it complaining about. That's what you need to wait. I don't know. Oh wait it does say create new volumes that instance for me didn't you just ask me, we were looking at just did that. Didn't it. Not just do that. Okay, let's try this again. Right. I don't mean it we skipped it. No, I see. Okay, so go ahead. What about create new volumes that instance. We don't see it the blower in the morning in the pink morning. Oh, okay. Oh yeah, we did. I thought we did this. We thought we did this. Okay. Are we being impatient. I don't know. Okay. It goes under advanced. Nothing. Okay. It should just say, you know what, let's just select all. Nope, that didn't do it. Do you have to set a max size, your max size is empty. Just so you know, um, how is it. Yeah. Okay, and it's got no name up at the top. I thought we did this. How is there not because we have an MF coffee. What were you going to do? logical volume set. Triple zero. We just did this. All right, now the next button's good. All right. Off we go. Okay. You won't be able to go back after this. You told me that last time. Okay, now here we are. See, no, look, we did do something before. Okay. And now it's asking us to do it. Create a new volume set instance. And now I'm getting, now I'm getting slack messages. This is good. I hope they are from our friend, Mr. Bloom. They are. Yeah, we're doing something wrong. He's like morons. Don't do it that way. Okay, hang on. I can see. This is his area of expertise. Okay, so he told me to open this. Do this. Okay. So he says there, you'll find a YAML that will use local disk directly without the need of LSO. Just change the path and the capacity. Where do I run this, Mr. Bloom? Oh, I just run against the cluster. I guess that's weird. Make sure I see status at the right cluster here. Yes. So I will need some sort of real quick. Okay. Pull it up over here. Edit this file. And it is doing work in the background. You see here, your nodes are increasing. It's checking them out. Is it? Yeah. Yeah, you see him. You just keep starting with one. Okay. Anyway, we should do a Chris sets, but, but just, well, yeah, but he says to do this. Okay. To bypass the local storage operator, which we already have installed, but okay. Chris, which one is better? For your particular setup. Notice to users, we don't normally run things this way. No, like, that's the problem. But let's just create. Okay. All right, apparently there's stuff. Okay. Things are changing here. It's doing its thing. Okay. Now we need to be really patient. Okay. So you, if you want to, you can do like OC get pods on the open shift storage name space. If you just want to watch stuff happening, but it should complete. All right. Okay. It's waiting for that progressing true. Available falls. It will, it takes a little bit. Yeah. It does. And what do you want to watch? I'm sorry. What did you say? Well, do you want to go to command line and, or you can go to pods, workloads pods, like either way, just to, you know, Yeah. Watch it in action. Yeah, it looks nice. Yeah. All right. Here we go. Yeah. And that's all of these have to do their thing. And hopefully no one's impending everyone will be completed or running or something. Yeah. Yeah. It's just setting itself up. So we're not, other than that part, we're not going to do anything special from the OCS side for quay. When's the last time you installed a registry? Have you, have you done it recently? We, I have not done it recently. It's been a while. Did it. I can't remember it was the last week of this week. All the weeks blur together nowadays. So sad. I think it was last week on the channel, we built out a quay code. No, yesterday. We did stuff with quay yesterday on the level of power, I think. So, Oh, cool. Okay. It was with quay, the service, not quay, the local registry. Oh, okay. Okay. Okay. All right. So we're going to do, we're going to install it. And then by default, it's going to use the new band points. For its object storage. There's, it does more than that. It does more than that. It does more than that. Like I think the Postgres or whatever database is using is, doesn't use that. It's like the registry part uses the new end point, but, um, But you'll, you'll see what we've got going on here. So. So yeah. Okay. The, the snippet that. Chris Blum sent. Said, you know, he said, you know, use that if local storage operator wasn't working, it appears to be working fine. So we're doing great. We're going to run with it. He also noted. The cluster is considered a compact because your nodes only have eight CPU units. Okay. That, that is interesting facts. Um, I would not think eight CPUs would be compact in any way, shape, or form, but okay, fair enough. And we all have our different definitions of compact. Well, you know, we're storage people. All the things. All the things. Right. No, I get it. We do actually expect. I mean, this is why like people have, this is why people build literal storage arrays, right? Like it is a unique thing. And we're doing it with commodity off the shelf. Consumer SSD. You know, so. Yeah. You know, regular Intel CPUs and stuff. So. Yeah. I mean, for testing around compact is fine. Right. Like it's for what we're doing. It's okay. It's just not what you want anywhere in your production or something. Yeah. No, you don't want to undersize your storage. This is literally just. This cluster is designed for this purpose to use on the show. And then I can rebuild it whenever I want. You know, that's like always a strange conversation on the OCS side. When you, you talk to potential customer and you're saying, well, you know, you need so much. It's, it always sounds like so much. I'm like, but, you know, you're kind of doing this anyway. Off with other kinds of storage too. You're always. I mean, what's the minimum amount that I could get away with to support these $1 billion devices. I just like, no, you need to do it. You always think over capacity one version means. Right, right. I did funny that like, for some reason in this context it seems odd, but I'm like, actually it's just kind of in line with. It's not odd. Right. It's odd because. We're used to like, you know, eight CPUs on a laptop is really good. You know, like that's why it's odd. Right. Because we've all been in this world of what hardware can I get my hands on. Not much. Yeah. But yeah. Right. They're still coming up. Okay. Okay. Have you ever looked at any of the. Horizontal pods scaling stuff. A little bit. Not much. I'm going through chat right now. Let's see. Okay, cool. So. Mannequin Dan wants a tiny demo lab environment, but OCS needs very huge multi-note setup. Like we have mentioned. Can we have an all in one OCS cluster just for experimenting? We cannot test OCS on CRC. Obviously. Is there a recommended alternative? I mean, this is kind of it, right? Like spending up multiple VMs in your VM environment. Yeah. Using some kind of storage that's already available. Yeah. Right. Mannequin Dan, if, if, if you're already an open shift customer, we can help you with that. So I would strongly encourage you to reach out to your red hat account manager. If you have one, if not, please feel free to email me and I'll get your questions to the right answer. Yeah. So that question is really how small can you go? Sorry. Right. Basically. Yeah. Okay. Just how small can you make this? And I think this cluster is about it. Right. Six control plane or three control plane nodes, three OCS sized minimally nodes, and then an extra worker nodes. So like that extra worker node is if we want to spin up a VM. Right. So I mean, maybe you can find out some unsupported stuff. If they're willing to, right? Like there's, there's like the official line that we've tested and written about. And then there's like, oh, you could try getting away some other things. So we can. Yeah. This is definitely the try to get away with that concept right here. Right. Like what we're doing. And then playing off Mannequin Dan's post. Is there another open shift live show that is open shift for dummies. And the door prize is a cookbook for starting yourself in the smallest possible cluster. Yeah. Yeah. Yeah. I mean, so there are a bunch of ebooks. There are a bunch of learning resources, which I'm dropping links and chat to right now. But John, specifically, is there an open shift for dummies kind of show? I would say. Not to insult Andrew in any way, shape or form, but Andrew Sullivan's. Open shift administrator office hours. If you're working on the open shift admin side, right? Like your, if your world is in this administrator view, which we're in right now, I would highly recommend joining Andrew's show. It's Wednesdays at 11. I'll drop a link in the calendar. Or drop link for the calendar. Sorry. And then if you want to get at it from the dev side on Tuesdays, 11 o'clock, there's the developer experience office hours. So if you live in the dev console, that's for you. That show on Tuesdays at 11. If you live in the admin console, the show on Wednesdays at 11. And then there's also the level up hour, which is like taking you from I'm a Linux admin to I'm in open shift admin, right? So that whole thing is that whole thing is a program. And the show is part of the program to help you level up your skill sets to get you to the point where you're administering your own open shift clusters, just like you're administering your own rail boxes right now. And John says, I'll take the admin side. Thanks. So yeah, that's Wednesdays at 11. I think every other Wednesday 11. I forget it's okay. That's why the calendar link is there folks. It'll keep me straight. So I think. Not quite yet. So the way this works, right? Like, let me get this straight. Seth, Rook, Nuba, that's how things get installed, right? The kind of Rook, Seth, okay. And then, and Nuba, I mean, so the like Rook and Nuba, I think actually attack things at the same time, but, but separate. Yeah. Yeah. So Seth. So Rook is the operator for Seth. Rook is the upstream. Okay. Got it. And Nuba is the operator for all the multi-cloud gateway stuff. So are we good? Not yet. I mean, this says one running, running, running, running, completed running, running. It looks good. As soon as those are like, they're marked as. What is the deal? Oh, so the OSF operator. I'm trying to see. So it won't be marked as done. Complete running. Sorry. Until everybody's happy. Right. So OCS, right? So you're saying, so OCS is up here and then Rook and Seth, and then Nuba and all the MCG stuff. So back up here. It's common though. It takes, it does take a while. So hopefully everything's running. Okay. Otherwise we'll go investigate and ask specialists around here. No problem. Okay. You're going to contain. We appreciate you watching. And if you have any other questions, feel free to let, let us know. Let's see. Remember for OCS your statements are correct, but for minimal OCP there is. Right. Correct Lloyd. CRC exists if you want to run minimal OCP, which is open shift container platform to run minimal OCS. You actually need. No, there was what wait, there was something else that wasn't quite that wasn't one of one, unless it was completing. No, sorry, my, the new the endpoint. It's not yet done. So it's not marked as. Thank you. Oh, sorry. Okay. So the refresh look good after your refresh. Okay, it does look good. Yeah. Yeah. No, everything smart. The OCS operators. Everybody's happy. Okay, cool. Let me know in chat if you can see this better than I can, but it looks good to me. So let's look at the OCS operator real quick. It does look good. It looks good. Events just to make sure. Writing this probe failed. I don't necessarily know. Yeah, it says status running. So I'm going to take that as a good thing. Always go look at the dashboards. initialized ready. Okay. Oh yeah, good point. We can look at the dashboards. You can I mean that's always like the quick way to is it under storage or is it in the. So I like the ones up at the top just as a quick check all the way at home. As a place to be like begin. Yeah, so you can look at persistent. Yeah, like everybody should be happy. Those other two other tabs come straight from OCS persistent storage and object storage up above next to cluster. Nice. Nice. They should look out click on something make sure it's okay and happy it should be. There you go. Okay. All right. Oh, and let's just look. Can you click on object service just for sure. Absolutely. That's you. That's your thing. Okay, so here we go. When to start with the red one. Let's start with the registry. Okay. All right. Okay. Let's go on. Yeah. Key. I'll say it the right way just for you. Thank you. Okay. Okay. No, no, no. That's the, that's the idea of the show. I confuse you and then you set me straight. All right. All right. So that's a clay registry. Yeah, we'll take the latest version. Sure. It's going to install on all namespaces. Okay. As opposed to just. Yeah. I guess I want that. Okay. I'm going to try the open chest storage namespace anyway, like, like later on when we do apply that deploys a billion. It's as long. Yeah. I don't know if it matters. So let's try it. Why not. Push comes to shove. I use the assistant installer for OCP and I build a new cluster. Getting very good at building new clusters from this box. Okay. So from what I remember, when we go in and do this, we're really taking all the defaults. It's going, it should pick up, it should use new before office storage for the part that needs, we don't actually change anything. Where I remember the. Do we need an active subscription for installing the quay operator. To my knowledge, yes. Do you have, but if you're using it and like CRC, you technically have a subscription if that makes sense, right? Okay. I know I haven't run into that. Yeah. I'm Manica. I'm wondering where you're running it from. But yes, you would need a subscription to open shift. In production, for example, this. Does the quay registry replace the open shift image? It doesn't replace anything. It adds to, right? So the, the internal image registry doesn't change. This is a service that you can expose for others to use. You can do it from an admin point of view. That's really nice. Right. Like you, you can split the two layers.