VMware TGI Kubernetes, 23 Oct 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: TGI Kubernetes 137: Waypoint

Description

Come hang out with Josh Rosso as he digs into Hashicorp's new Waypoint project! As usually we'll talk about what's going on in the cloud native space, then transition to playing with Waypoint.

00:00:00 - Welcome to TGIK!
00:03:53 - Week in Review
00:18:30 - Waypoint & k8s Abstractions
00:35:01 - Installing Waypoint and Doing a Deployment
01:01:21 - Installing Metal LB to enable Waypoint server
01:14:20 - Using Waypoint Server and doing more deploys!
01:49:11 - Wrap-up and Goodbye!

github: https://github.com/vmware-tanzu/tgik/tree/master/episodes/137

A

Hey tgik how's it going happy friday happy friday from uh kind of cold colorado. Today I don't know if any of you out west are also feeling some of this, uh this cold weather, but it's like 10 degrees in the mountains we got some snow seasons are changing out here. So it's exciting times all right. Who do we got joining us today, flash how you doing flash swarner from hashicorp thanks for joining us swarner you can.

A

You can help us out if we uh say anything silly today we really appreciate you jumping on so short notice. What's up maddie, what's up martin great to see you too, as always olav noel, eric, hey eric hope, you're doing well, choco nice to see you as always haim great to see you tuning in from israel. I'm glad people make this time slot work. I know it's kind of a weird and tough one. We got jose coming in from panama, hey, jose.

A

We've got riv from boston, hey riv, all right, hey, rich glad! You can make it thanks! So much for joining um rodolfo from texas, beaujonce, hello, hello, hey noelle, and we've got craig from the midwest. What's going on craig, I'm from uh I'm from cleveland ohio, so I'm a midwestern person too. I don't live there now, but I used to live in the midwest, elise pavel from bulgaria we got joshua, hey josh um got cody from hashicorp 2. thanks for joining us cody. We appreciate you being here: elko, hello, hello, perusan, rada from scottsdale.

A

We got preethi from bellevue wow. We got folks from all over today, richard from austin savvy thanks for joining us savvy awesome awesome. All right, oh marcin from poland. Great to see you thanks so much for joining in aristeads. Thank you for joining as well. Hey, hey, borco, um great, to see you yatin from virginia all right, so we gotta. We got a good crew so far, um thanks so much for joining us. If this is your first time on tgik, I see a bunch of regulars, but maybe some folks.

A

First time, tgik is a weekly episode that we do on fridays at 1 pm, pacific time. The youtube time is oftentimes configured wrong because of time zones. So hopefully that didn't confuse any of you. Nonetheless, today's episode is on waypoint. It's our 137th episode by the way and we're going to be digging into this cool new tool from hashicorp.

A

That's getting a lot of buzz and uh you know in true tgik fashion. We figured why not just learn it uh on a stream with y'all, so we can kind of all all learn something. So this would be. This would be a really cool one now tgik, along with jumping into a piece of tech.

A

As many of you know, we like to start the week off by covering a little bit around, what's going on in the cloud native ecosystem, so we'll go ahead and start there and talk a bit about what's going on in the kubernetes space, all right, who else we got joining joe thompson, hey joe uh funny story about joe. He and I worked together at core os a long time ago. It feels like it's been a long time ago at this point, um so it's so great to see you joe, we don't.

A

We don't cross paths, often enough, especially now that uh that cubecon is not really a thing anymore. At least sorry, the in-person kubecon's not really a thing um cool, so we can review. What do we got going on kubernetes 120 alpha 3 is out. You should check it out if you're contributing to the code base do note that november 12th is an upcoming code freeze and if you don't already have it bookmarked and you're involved in release stuff or just want to know.

A

I guess really just what's going on release wise, you should bookmark this page. It's very conveniently kate's dot, dev forward, slash release gives you some great information. It's probably something. I should be visiting a little bit more and checking out like the enhancements and all that good stuff, but it's very cleanly laid out. I think it does a really good job of succinctly articulating. What's going on so we're driving towards 120 getting closer and closer check out the alpha.

A

If you haven't already and that's kind of all, we got oh yeah speaking of coros, I should have mentioned I'm wearing my rocket shirt today, because somebody, I think somebody last week said they had a rocket shirt or something. I can't remember what we were talking about, but I am uh I'm wearing my rocket shirt as well and representing so all right, cool cool, so noel, you said coarse was the first vm uh cloud vm you spun up and you didn't understand a thing at first.

A

Well, that's an interesting entry point for your first cloud vm, especially given, like all the opinions and all of the kind of focus for running containers that the core os distro had in it. That's pretty cool um tim you've only got a rocket sticker yeah. Maybe maybe we can reproduce these someday as like a retro shirt, that'd be kind of cool um all right. So what do we got going on in the ecosystem this week? Anyone here use fzf uh fuzzy. Is that how you say it? um I I use it in vim.

A

I forget to use it in bash all the time, but it's a really cool tool does uh fuzzy searches of things and it like plugs into freaking. Everything like you can have it in your in vim, open up a buffer in vim you're searching through code. It's super fast. It's super rad. It can do your bash history if you're a control r and do the whole reverse search thing and then brad borrows. Hopefully I'm saying your name right.

A

Brad has a new post about aliases and functions using fcf for kubernetes and what's really cool about it. I have not tried it, but I think this is how it's working is when he runs the command he's actually like scrolling through or fuzzy searching through pods, and then just the described data is coming up in a buffer to the side, so that is pretty freaking cool. uh It looks like it's a bunch of bash, um but I guess it's kind of cool like you know, just some bash aliases, not another binary.

A

You need to install or anything like that. So this will be definitely on my radar um in true form. For myself, though I'll probably put all these bash scripts in place and have all these cool conveniences and then forget to use them because my muscle memory always takes over. But if you use fcf fuzzy, I'm just gonna call it fuzzy um check. This out looks really cool. I think it's pretty sweet.

A

All right cool, alex you're joining us from uh gorgeous northern california, thanks for joining us, alex welcome, welcome all right cool cool. So what else do we got going on in the ecosystem? uh Learn kate's. I don't know if you all follow this site. um They've got a lot of really great educational information on kubernetes. I think they might even do some consulting excuse me if I'm totally off on that. Nonetheless, uh they put together a google doc that does a side-by-side comparison of ingress controllers.

A

Pretty cool, since uh our world today is uh filled with many many options for ingress uh kind of cool. They went by protocols by clients by traffic, routing capabilities, all that good stuff, and you can just kind of scroll horizontally here and see what nginx offers kong traffic, h.a proxy istio's gateway, contour on and on and on pretty cool. I feel like we need one of these for uh pretty much every piece of technology in kubernetes. I think we need a service mesh one. We need an ingress controller, one.

A

We use cni one, and I know some of that stuff's out there in blogs and stuff, but this is really cool. I think it's laid out really interesting. um So if you haven't checked this out um check it out, see what's going on with ingress controllers, maybe it'll give you an idea of a new one to try out that you uh have yet to play around with so pretty cool stuff. Thanks, learn, kate's for putting that good info out and and all the awesome info that you all put together. I think they have like a.

A

uh I can't remember if they called it like a production, ready checklist or something like that, and it was just like kind of an interactive checklist. You could go through about how to harden your clusters for more realistic use cases um so uh or not realistic use cases. I should say production use cases, so they put out some good content, and this is just another one of those awesome things all right. Good deal so um skim looks awesome. huh What is skim? Let me check skim out. Real quick is skim, another search thing.

A

What's going on here, skim, let's see, half of our life is spent on navigation. It's probably not incorrect. huh I have to check this out a single executable, sk cool. huh So is skim a skim, a fcf replacement written in rust is that kind of the deal interesting stuff, huh all right, cool, so uh ops, psas things that we should probably be aware of.

A

You've heard this from us quite a bit at this point, but if you're using home v2 don't do that uh consider um upgrading helm, I think there's lots and lots of good resources going out right now. I've been seeing a bunch of folks from the helm, team and stuff give a lot of good info, so helm, repo in v2 end of support.

A

um This not only could impact where you're pulling home charts from and all that good stuff, but also is the end of life for v2, so like important security patches and all that stuff, like it's time to get rid of tiller and move on to the lovely lovely world of helm, v3 and all the cool stuff it can offer so make sure this is on your radar folks, if you're at all, uh still in the world of uh you know using helm for deployments, um just make sure you're you're got an upgrade plan here, you're aware of the deprecations, and you don't wake up one morning and go.

A

Why is everything broken or even worse? A cve comes out, and you say: oh my gosh. We can't patch this. We now are forced for uh helm, v, helm, v3 and, as eric put it I think, beautifully friends, don't let friends use helm v2. So let's get our! Let's get our friends on home, v3 how's that sound all right, cool cool, all right and wally.

A

You said you added, learn, kate's telegram channel to notes thanks so much waleed so check those out um right at the bottom here and we'll uh we'll we'll take we'll give that a click too in a second here all right. So uh it's, oh, I almost missed one. I almost missed one docker hub folks who use docker hub for images just be aware that some of the uh tos is changing. This post talks about it.

A

So if you currently pull from docker hub- and it's not just you playing around- um do check out the blog, this bulleted list, I think pretty succinctly- puts it. There's implications around the amount of pools for uh anonymous users and authenticated users in your free plans. Pro and team plans are not a problem and then tell me if I'm tell me if I'm forgetting here, everyone, but there's also some implications on like duration, an image lives in docker hub now, if it's not accessed, is that correct someone?

A

Someone correct me if I'm wrong there, but be advised how docker hubs, uh approach for kind of the free plans go? Is changing a little bit pro plan team plan? All those things just be advised of that and again one of those things don't wake up in the morning and wonder why your ci jobs that pull a bunch of stuff are suddenly just exploding, keep an eye on it and make sure you're either paying docker or paying docker hub.

A

Whoever owns it now, marantis, right and uh and or you have an alternative registry, if you're not going to continue with docker hub, so is it six months savvy? Does that sound about right? Possibly so all right so wally, you said images get deleted if they are not accessed over a certain duration. Yeah! That's what I was thinking so like you know. You probably could work around this.

A

I'm sure there's already been some clever folks who have written, like you know, a thing that pulls the image every now and then, but you know maybe just consider not if it's, if it's not just for fun, maybe consider doing something: that's not that big of a hack because you'd hate for that to fall over and stuff to blow up on you, um okay, so that's docker hub, that's helm, v2 uh be sure to go down, go up to v3 on helm, cube academy, the asking anything with joe beta.

A

I think you all know joe peter very well he's the um the common host on tgik and has bootstrapped a lot of this good stuff. He didn't ask me anything on cube academy and there's a recording available. So if you want to check that out on cube academy click, the link in the notes go to the ask me anything and you can watch the webinar good stuff all right and uh alex says what the stuff want to be around for centuries. That's a good question alex. I think the answer is we're.

A

Gonna put it inside of a frozen igloo thingy inside of the arctic. Just like github did for their uh for their arctic uh code storage right. We need a. We need a docker hub arctic uh code storage. Could you imagine the the junk of layers that would be in that that freaking thing? I think archaeologists would just be like what were these people thinking holy crap anyways other stuff, that we've got going on hands on cka ckad with a with a cks flavor. So I haven't actually seen this one.

A

This must have been a later edition. um Oh waleed sweet hey, so um you should check this out. It looks like it's something: that's going over cka ckad um for some reason: okay, cks! Is that the is that the security certification? I'm sorry that I'm blanking on that. um For some reason, cks isn't uh resonating, but I'm guessing. That's the certified kubernetes security exam sound right check this thing out, regardless waleed rocks. You should check it out. It's probably going to be great uh security.

A

Specialist good deal so yeah give it give it a give it a give it a view and uh awesome wallet, I'm glad to see that you're you're throwing some content out there like this. This will be uh hot stuff, especially with uh with people looking to cks awesome, good, good, good and, as choco says leet. I totally agree very, very elite of you all right. uh Oh okay, cool, there's, a uh github repo as well. So do note. We have both of these linked github repo.

A

The live stream is up there as well check those out put them on your calendar, all right. So a couple more items for this week we've got cube cuddle aliases, so I have also not looked at this one. This was probably a recent addition, but this looks like it is: ah okay, some some example aliases that make cube cuddle more convenient.

A

I feel, like you all, are calling me out here because, as you know, I I type out every command like very verbosely and very anti-aliased, so you're, probably just like josh, just make a freaking alias already. We are so sick of watching you type cube cuddle and I feel your pain. The problem is, I make the alias, then I forget to use the alias but yeah. This is really cool. um 800, so generated aliases check these out. It looks like it must be something that actually auto gens them, perhaps so bringing some things together.

A

I'm trying to parse like k, sis gpo so kubernetes. ah I see I see, okay, kubernetes uh cube system, namespace, get for g and then pod, that's kind of cool. So it's just kind of like a convention-based thing where it's actually kind of bringing up your aliases uh automatically, which is yeah. I can kind of get behind that man 800 aliases.

A

That kind of scares me- I don't know if I can handle it quite yet all right, cool cool, hey, fully geared bear glad to have you, as always rex you're on the team, anti-alias yeah, I just I just. I know that they would save me time. I just cannot bite the bullet for some reason.

A

All right, cool joe says, alias k equals cube cuddle. At this point he refused yeah. I hear you we're sticking with the cube cuddle, the full command all right. Last thing: we've got here um that I think waleed added in uh telegram learn kate's. So I have not used or looked at telegram. Yet I'm guessing we haven't done what even is telegram. I have no idea.

A

Oh wait did I say, learn kate's. What am I thinking? Wait? Learn kate's, apply telegram, okay, what is what's going on with telegram someone's gonna have to give me a a quick uh overview of what the heck's going on. Is this an app that the learn kate's group has built? It's an iphone app? It's it's running on cube. I assume I'm just trying to figure out. What's the tie between uh learn, kate's and telegram, I'm under a rock with this one?

A

ah Okay, it's such it's an app like whatsapp and uh and signal. I see. Okay, it's a chat group got it all right, connect the dots for me, though, what's the relevance to learn case, are they the ones who made this thing? Is that? Oh, I see sorry sometimes sometimes it takes me a little while people, okay, it's a group on telegram around learn. Kate's. I see okay, totally grok. What you're saying now? Sorry, I just you know I just I just had to connect the dots. No big deal.

A

Yep zavin says the world needs more chat, platforms, yeah you're, not you're, not wrong, and the funny thing is: we've got so many chat platforms and the majority of the world's probably using unencrypted uh sms right.

A

Oh man, okay and tests get to work, cool cool all right. So those are our notes for the week. I think we're about ready to get into waypoint some pretty interesting stuff. It's a little bit of a light week, but hey it's uh it's! It's got some cool stuff going on. So let's talk about wave waypoint, all right people in the chat- um I guess the maybe the hashicorp people are a little bit exempt from this question, but you're welcome to respond anyways. I think I know the answer who has played with waypoint.

A

Has anyone gotten the chance to play with it say yes and chat if, if yes and and if you haven't played with waypoint yet say no, I'm super curious. I feel like it. uh It got announced recently it's been on our radar um to check out. We just haven't gotten the opportunity. Has anyone gotten to check it out in chat? Yet?

A

Okay? Yes, yes, cool! No well hey, I I appreciate the nose. It just means you're waiting for tgik to uh to check it out cool yeah. It seems I agree. Well, it seems like a super super interesting concept and I think we should, as usual, kind of uh yeah cody you're, obliged to say yes yeah, you wouldn't want to go on record saying no there.

A

um So this is a super interesting space and I'm pretty I'm pretty passionate about this space and feel really really intrigued by waypoint, because here's here's kind of the thing in my eyes right: um let's, let's, let's open up a mirror and this way we can have some uh some diagramming as we go and learn a bit about how this tool works right. So, like here's, here's how I kind of view the world today right. So we right uh the tgik community, the kubernetes community and all that good stuff right.

A

We um oh, is there not a heart there we go! That's what I'm looking for. We love.

A

Oh there's a kubernetes logo. How cool we freaking love kubernetes right I mean I, I think, there's no question most of the people in this chat, assuming it hasn't fallen over for us and we're losing our weekend because we're trying to debug our clusters.

A

I think, generally, we really really like kubernetes, or else we wouldn't all be here- and I think what's really interesting about that- and I see this a lot when I'm helping folks kind of drive towards production with kubernetes is while we might think it's the best thing since sliced bread, there's some merit to the fact that a lot of our downstream consumers clients, if you will might not really love kubernetes the same way. We do.

A

They probably don't hate kubernetes right, but their priority is a lot more about shipping software effectively and and the the means that kind of gets them to that point. That efficiency is really the crux that they're concerned about. So when we go to them and we're like. Oh my gosh look at all this cool stuff, you can do with a yaml file and there's this reconciliator and this really cool scheduler.

A

They might be like yeah, like that's cool, but at the end of the day, I don't think it's at the same level that, like we, think all these internals and all these bits are just super super awesome right so and I'm not trying to say cube, isn't awesome. I'm just saying we have different sets of concerns and a lot of us in the kubernetes space.

A

What we're really trying to do right, we're not just like um inheriting kubernetes for the sake of inheriting kubernetes, although maybe, if we're doing like some resume driven development, that's our intention, but more so we view kubernetes as kind of a larger piece of the puzzle. We view it as something that if we bring kubernetes down- and we think a little bit about some of our underlying pieces like back in the day, we used to always talk, and I guess we still do talk about the idea of an ias provider right.

A

So this could be- and this is not just a cloud provider right. This could be amazon, but it also could be your on-prem uh kvm stack. You know whatever, whatever you're using right, and we kind of thought of this as like all right. So, let's abstract this thing now again kubernetes itself like at its core. What what does it do like when you think about kubernetes at its core?

A

What are the main things it solves for in my eyes, right, I think of kubernetes as a thing that moves containers around hosts and don't hate me for over simplifying that like it, it definitely has like some interesting primitives for the service. Abstraction and cube proxy is technically part of the core. So technically it does some weird mangling of ip tables and stuff. To give you some other abstractions, it gives you a really robust, declarative api, a controller and reconciliation model.

A

But you know the end result of those pieces put together is a container orchestrator right. So when we take that a step further oftentimes, what we're kind of thinking about is this notion of all right?

A

What what when I have customers, what the heck do they actually care about, and how do I provide the things they need so that a developer isn't just being talked into learning kubernetes, but they're actually gaining value so a lot of times we throw these abstractions on top right, and these could be all kinds of things like I mean since we're in hashicorp land.

A

Let's talk about vault, real, quick one of those really common, really freaking cool things is vault rather than having application developers go in and um you know solve their secret management uh secret management component 30 times over in every single app with 90 different services we can put vault in kubernetes.

A

We can integrate it with the injection sidecar mechanism that vault offers, or if someday it stabilizes, the um csi driver that can go out to vault and grab secrets and, in short, at the end of the day, for an app, they can just go in and say, give me the secret and boom in their file system. They've got their secret, that's an amazing platform service that again provides value, and then it just continues going up and going up and we have, like you know our load balancers through ingress controllers.

A

We've got our, um you know, observability stack, so maybe we give out like uh self-service, uh prometheus and all that kind of stuff, so you kind of get where I'm going here. It's about the bigger picture now the interesting thing kind of becomes in in my mind- and I think this is you know- maybe where waypoint uh waypoint hits a really or scratches a really interesting itch is when you've got all these things in place.

A

There's kind of this inherent question of what's the right abstraction for these things, that we've got at least at least this is how I reason with it in my head right. So let's just put: uh let's put people to the side. These will be our our developers here. um Okay, so when we go in and we kind of think about this it, what we're really thinking about with the abstraction is what's the right level of abstraction for folks to be successful. With this platform we build, I mean kubernetes are not.

A

This is a universally true concept right and it might seem, like you know at first, maybe there's a right answer out of the gate and I'm I'm a bit on a soapbox right now. So I apologize but oftentimes. It feels like you know, someone will send a tweet out and they'll say something like if your developers know they're using kubernetes. You failed, and I totally disagree with that. I think that's like so over the top right.

A

I think the sentiment could be right in certain cases, but I largely think of this abstraction problem as a bit of a spectrum right, it's kind of the spectrum where, if we think about the abstraction of kubernetes or any any platform, this is this is more than just kubernetes. This is the platform as a whole, but we'll use kubernetes as the example there's really a whole a whole spectrum.

A

Here, on the very left side, we could expose raw kubernetes like, for example, let's say that we're in a model where we just want developers to sign up on eks spin up their own clusters and have at it do your own thing. We don't we don't we're, not worried about multi-tenancy and all that you just you, have your own cluster go for it now a lot of times.

A

That's a terrible idea right, it's totally terrible and on top of that, in a lot of orgs, if we gave orgs raw kubernetes, so I've like I've done a bunch with like telcos and banks. At this point, if I went in there and said we're, gonna bring kubernetes in and all your developers need to learn kubernetes that would absolutely not fly and a lot of times in kubernetes land. We kind of think about it like oh well. What do you mean? They can't just learn like a yaml file, but it's way more than that right.

A

You need to learn what a replica set is. You need to know the relation between services. You need to know how to wire up x, y and z. It's such a conceptual overhead, like learning, kubernetes sure the interface is this declarative yaml file, but making correct decisions and doing things effectively in the system it's more complex than that. It's it's freaking hard. um It is. It is really hard, especially when it's not your world and then on the other side of the spectrum. We have what I would call the full abstraction.

A

Now I think waypoint will maybe play a little bit closer on this side of this spectrum. But let me let me talk a bit about the full abstraction for a moment when I think of a full abstraction. I don't know about y'all, but I'm thinking about like heroku, I'm thinking about cloud foundry, I'm thinking about things that, like, let's think about heroku, so everyone in chat um say say: yes, if you've used heroku before I'm super curious, does anyone?

A

Let me let me just pull the audience who here has used heroku before to any extent, give me a yes, even if you just played around who's used heroku, I'm seeing a couple. Yeses come in. So, okay couple a couple: folks: yeah: okay, good good. So what do we know about heroku? We oftentimes refer to it as like a platform as a service, and I know this is kind of one of the things that waypoint is uh kind of approaching it from a new lens or a new angle. Perhaps um what's what's interesting about heroku?

A

Well, you kind of just like take your code base and run a command, and I admit I'm oversimplifying it but like somehow it just shows up in a cluster. You get an endpoint and then boom and think about that from a developer. Experience standpoint right, there's, there's tons of trade-offs with this approach. So I'm not trying to say like heroku, is a golden state there's a lot of reasons why it actually might be terrible for you, but it is.

A

If, if the model and opinions of heroku work view, it could be an amazing model, you run this command. Your app starts running. It went through some weird build pack thing that you don't even have to really worry about and then boom you've got a url that you can start hitting. Does heroku use kubernetes under the hood, who cares like who? Who cares if they use kubernetes under the hood, because that level of abstraction is something that prevents us from being worried about some of those pieces?

A

So what I'm really trying to get at here with my with a bit of my soapbox, is how much we should abstract kubernetes is really nuanced and honestly, I think the best way to make this decision organizationally is to talk to your end. Users, don't just go, build a system and abstraction and not bring your developers along for the ride. Your developers for a lot of us at least, are our customers right and we want to make sure the abstractions we're making are right, because it's not just raw case or full abstraction.

A

Either it's not a binary decision. You can have entry points that are kind of in the middle things where you have a happy path, perhaps through something like waypoint. I don't know will find out today, but then also expose the underlying knobs in certain respects that don't don't put your system at risk and other things like that. So it's a really really interesting thing to think about. um Pierre said netlify might be a better yeah. Exactly.

A

I think netlify is another awesome example: it's really that kind of end-to-end experience and how do you build it so with all of that ranting in place right? Let's, let's talk a little bit about um about uh waypoint and how it provides um an abstraction like this, and it's actually a pretty unique approach to thinking about this abstraction right. um Oh hey, steve yeah!

A

I was just talking about the the rocket shirt too, so we're uh we're on the same page, we've so far in chat, we've got a rocket shirt, a rocket sticker and uh there's lots of rockets going on today, all right, so all right, waypoint waypoint, is a project that got announced by hashicorp. um It is super super interesting and there's there's three things that from what I understand about it there's three things that it's focused on.

A

um It seems it seems like it's focused on bringing a path like experience and hashicorp folks that are in the chat. You are super welcome to correct me. If I say something silly, I will take no offense. Okay. First time, looking at this thing, it can theoretically bring a path like experience by focusing on how to solve the build, deploy and release components that come along with the paths.

A

Now I think, what's what's pretty interesting about uh waypoint, at least to my understanding so far, and we might, we might learn differently here pretty soon, but I think what's kind of interesting about it is when you think about this idea of deploy when you think about this idea of um sorry, if I'm using the wrong terms here, we'll learn them soon roll out and then, if you think about the idea of build a lot of things that you end up with with these passes is like, if you're in heroku land or cloud foundryland, they're defined, it's a good user experience, but like it just is what it is and what's really interesting about waypoint, at least from what I understand so far is we can basically plug these in to different components that make sense so think about the idea of providing an experience um somewhat.

A

Let's say like with one layer of indirection from what the underlying things are that satisfy it. If that makes sense so like, let's make this really concrete real quick, what do we mean for build? Well, maybe you want to use uh build packs right um so build pack. Maybe you want to use um docker files for build so we'll put in oops, we'll put in the docker file here right, so we've got build packs.

A

We've got docker files and I think one of the interesting ideas here unless I've got it totally wrong, is that by running a build command of some sort, we'll learn about how you do it in hcl and all that stuff soon, these details, don't necessarily have to be front of mind in a lot of ways. We can really kind of run something we can plug into some provider, and I would imagine waypoint is super extensible for defining custom functionality.

A

I'm sure that's going to be a crux for a lot of you know large enterprise users and then same idea for things like deploy. What, if our deploy experience, is consistent for ecs in the same way that it is for kubernetes right and from the higher level, you know abstracted view of our end user. What? If we don't need to worry about the intricacies, the underlying complexities, the deltas between these things, we just know how to use up or whatever the command ends up being with waypoint, to make these things happen right.

A

So super super interesting stuff, and I think this is a. I think it's a pretty interesting, take on the paths space. That's why I'm so psyched to check it out with you all today um joe says he has his core os challenge. I know there's so much core os going on lately. I probably gotta dial it back a bit. um uh Where does mattie asked? Where does waypoint fall on the stability spectrum? It looks like you've got some answers there, maddie so a point one release so correct me.

A

If I'm wrong, it sounds like it's still early days, perhaps, but it's looking really cool and probably a good time to start uh start testing it out. Oh, that's interesting, noelle any hashi product in point. One uh is production, ready, huh interesting, stuff, um cool cool and richard you said planning to try and make a pack plus uh cloud native, build pack extension or cnab extension, um some limitation, but still a lot of power, very cool, okay, very, very cool um and uh walid.

A

You said you will still need to provision eks and not ecs can can you I think it can. Can it talk to ecs, you all have a provider already that talks to ecs. I don't know um interesting stuff, okay and cody- says waypoint isn't really made for prod at this point, which is fair, but I think, like I mean, what's what's really cool about this and I haven't looked at the provider or plug-in model, but the concept is really sound so like.

A

If the way you implement the plugin is open and you can do it, there's probably nothing preventing you from running with kind of these underlying providers and sorry, if I'm screwing up the verbiage, you all use at hashicorp, but I think of them like plugins or providers that could satisfy these types of commands. Now, if there's like major api changes in core waypoint, that's of course a different story, but it's it's just a really interesting. Take I think, on on the level this paz can play um all right. Oh no, steve! You got an incident.

A

Okay! Well, good luck! Steve we'll be thinking of you! We'll see you next week. Maybe okay have a good night, all right, we're giving you hug ops! Okay! So let's see if we can get waypoint set up here. um Oh here's, the answer to my question: aws ec2 and ecs. So it does look like it supports more than just cube. um Let's see if we can get this thing rolling.

A

So we'll start with uh the installation here, um manual installation, echo path: uh oh there's a binary okay, so we did open service mesh last week, like a lot of you saw and it looks like waypoint has a cli too.

A

I wonder if this id well, I guess waypoint having a cli, makes a ton of sense right. It probably wouldn't make sense for it to not have a cli. Given that end users are probably going to use it. So all right, let's get the waypoint cli in and start playing with this um I'm going to see real quick if I can download this package really easily. So let me just do a quick search. um It looks like they've got yum uh apt and homebrew and chocolate.

A

However, the heck you say that microsoft one, um so you can grab the the binary off of those. um What am I looking for waypoint right, waypoint bin? No, no, no waypoint bin, okay and uh .13. That sounds like the version y'all were saying in chat, so I'm just going to assume that's good. Let me go ahead and get this thing. Bootstrapped real, quick, so go in here, get clone waypoint bin all right and then I will make that package.

A

Lovely. Okay, so waypoint is being installed on my system and if all goes well, the binary should show up wally. You said the cli ui is awesome, interesting! Okay! Let's, let's check that out. So I think if you're saying ui, you mean there's a dashboard that the cli will pop up. If so, that's pretty cool. Okay, let's see if we got waypoint now hey it's waypoint! Okay, great! That was easy enough all right. So what do we got going on with waypoint?

A

We've got build, deploy, release up um okay, so this kind of looks very, um very uh consistent with hashicorp stuff, I'm seeing up right, so you probably can knit in oh and it's down here yep. um Oh, this will be an interesting one for us, so there's an install command, install waypoint server to kubernetes, nomad or docker. You know, I should have made sure that my I used cluster api to spin up a cluster before this episode and okay, it's healthy good. um I didn't double check it.

A

Okay, so we've got that we're looking good, um I'm guessing we're! Probably gonna start here. So let's go back to the docs and see see what we got going on. um Okay, so we've got that validate the installation. Okay, we know that that's good. It looks like they had a kubernetes one. Let me let me go to the docs real quick, actually, because I was looking through these earlier and I wanted to pull up the architecture with y'all, um so we got the cli so far right and I've got a kubernetes cluster.

A

It looks like there's going to be some type of server component. We're going to be running so server is the waypoint server. It's long running, it's a central service that serves the api. The api is consumed by the cli, that makes sense, entry point and other consumers, so I'm guessing server is going to go well. I guess it's actually telling me right here. It's going to go in my cube cluster or nomad or amazon ec2 somewhere, and then I'm going to be directly interacting with that server is what it looks like.

A

We've got an entry point which is designed to not affect the child process of any failures. Interesting huh I'll have to take a closer look at that and then runner, which is dedicated runner. Oh, if the dedicate runner is down the capacity server execute remote operations is diminished. Okay, I think that's just some more internal stuff, so we should just have to get the server installed. Let's see, if I go back to the uh the github page, I saw a link for cube, so I'll stay stay a bit on the happy path.

A

Initially at least okay um and wally. You said you can do a docker, okay cool. I do have a cube cluster, so we'll probably start with that, and then maybe we could do um a cube cluster and a docker uh docker daemon, hookup side by side, that'd be kind of cool all right, so cody says a ton of work on. It was with kind and early testing cool cool.

A

So we should hopefully our cluster api cluster won't cause us any issues, but if it does no beggie we'll figure it out all right. So let's go ahead and clone the repo. It looks like that's the first thing they want us to do so we will go in to the development. Actually, sorry, let's go into my there. We go my tgik episode just in case. We want to commit this. For any reason.

A

I doubt we will so we'll bring in waypoint examples get and we're good there now and if we go back over, looks like there's a node.js app, we'll be working off of there's kind, creating the cluster. I've already done that and here's what I was looking for: okay, install the server to kubernetes, all right cool. So, let's grab that real, quick and we'll take a take, a quick look at what this does so. First things. First, as usual, let's look at what I've got in the cluster. So if we do a cube, cuddle get pods.

A

um Let's see here, we'll just do for cube. That will do for all just so you can see what I got going on so make that bigger do that there we go okay, pretty normal cluster right. We got a cni, we got fcd, we've got cube proxy and then we've got some vsphere junk since I'm running all this stuff. On my on my home lab. So all that's fine and dandy.

A

uh Let's go ahead and put up a watch here, so we will set a watch again for that and then I'm gonna go ahead and run waypoint waypoint install and then it was what was it platform kubernetes except tos, so platform kubernetes and then we will accept tos cool, looks good, yeah looks good, let's try it waypoint install it's creating some stuff.

A

All right! We've got a waypoint server coming up.

A

Waiting for kubernetes service, okay, so far, only one component. So this is that server thing we looked at in the architecture that kind of makes sense and I'd. Imagine once we're up and healthy.

A

We should be good to uh to start interacting. With this thing from the cli um cody said, the cli needs to be able to reach the server and the deployed workload needs to be able to reach the server as well. Okay, we can definitely definitely can make that happen.

A

um Server has to be accessible from the cluster to have the endpoints working yep, we'll make sure we we do. Some routing here seems like it's having trouble starting so before we worry about uh routability here. Let's take a quick look, so cube cuddle get um we'll describe this pod. What do you say? Describe pod?

A

um Oh, no volume, binding, okay, no big deal, so the irony here is that I've got a csi running, but I have no idea if it's configured correctly, but we can work around this no problem. So here's what we're going to do, let's figure out what the pvc is right, so keep cuddle gap.

A

Pvc, it might be cool um for for waypoint um kind of, like I think, vault does this with their helm chart where you can run it in like a dev mode, and it won't require a pv that could be kind of a cool, uh a cool feature.

A

I don't know if that makes any sense at all, but pvc we've got this here, so datawaypoint server. Okay, we just need to make a pv match here, no big deal, so uh let's go ahead and do that kubernetes, uh persistent volumes, we'll grab a local volume and I'll be I'll, be bad and just uh bind to something that I shouldn't bind it to uh so we've got. uh Let's see, I need the. I need the claim types volumes. This is it right yep this?

A

Is it so we'll go to local and we'll grab a persistent volume right here? Okay,.

A

Yeah and joe says the vault helm chart runs by default is a three pod stateful set, but in dev mode yeah you can you can keep it. uh I'm thinking you can just actually, I know for facts. I've installed it a couple times without that, so that could be kind of cool, I'm guessing kind. I don't. I haven't even looked at kind recently, but I'm guessing it maybe like has a csi driver that does ephemeral allocations. Is that right, everyone just to kind of satisfy pv pvcs out of the gate?

A

So, okay, no big deal, let's get a pv set up, so we'll do uh pv.yaml um and uh we're gonna be really bad here. We're gonna mount to mount um and uh just don't tell anyone. I did that and now we just gotta figure out the host name, so cube cuddle, get nodes, uh we'll do a wide command here and I'll pick one of my hosts, not the control planes, let's roll with uh 6 5d, everyone, don't uh don't forget 65d!

A

I might oh they're all six five d. I should have said two: nine f, f. Five, two nine f is what we're going to use. Okay, so we'll go ahead and put that there we'll do that we'll say it's 100 gigs doesn't really matter. Persistent volume delete no storage class in this because I think the pvc was not using a storage class and uh let's do it okay, so we do a cube cuddle. We we're gonna watch our lovely pvc, so get pvc oops.

A

Maybe I should type keep cuddle. huh What do you think cube cuddle? This is why you all want me to have those aliases, so bad, okay, so cuddle apply. Pvml pv is created. Okay. Now, as long as I didn't screw, the pv up royally um the pvc should bind to it and then I'm thinking waypoint will start for us at least that's my hope, um cute cuddle. Oh there, it is it's bound, okay, lovely!

A

So obviously I'm doing something that would be atrocious practice in real life. But hey we're out here we're doing our thing in tgik cube cuddle, get um get pods! Oh hey check it out, waypoint starting up okay, so we've got a waypoint server.

A

What else are y'all talking about um cody, said ah kind in metal lb, so cody, I'm guessing, there's some kind of, like probably in the in the later like roll out phase or whatever you probably have some load balancer integration um with like metal, lb or probably any load balancer in cube, I'm guessing all right, waypoint's good. We should be able to talk to it. I think maybe possibly, let's see if our install is gonna come ready. Oh, you know what it's okay. So now it's reporting ready, but the service needs to be routable.

A

um Okay, I'm gonna do some really bad stuff. Today, I'm sorry in advance to everyone. So if we do cube cuddle get service, um the service is this right here? Oh wait! It's it's routable! On a it's routable on a node port, though, because it's using a load, balancer integration. I would think that that would just work unless it's waiting for something else, huh 10 interesting.

A

Maybe it's. uh We probably wouldn't know the host node port to go to. I wonder if it's trying to go to the cluster ip here we can. We can do something really bad, let's try it. um Why use? uh Why use a proxy when you can just change your routing tables? So uh here's our service ip um here is our. If we do a cube cuddle get pods for o wide.

A

We've got uh we're running on this ip address and then watch I'm going to bork up my routing tables. Then I'm going to lose you all on the stream.

A

I can guarantee it okay. So if we put the ip right here in my routing table, okay, let's see if this works invalid gateway. What did I do wrong dab via? It needs to be routable from your machine yeah. I think I can make it routable from my machine. um Oh wait! What am I doing? This is the gateway.

A

This is the gateway. Add route via 100. Wait, cube cuddle, get nodes, oh yeah! Silly me sorry hold on one second cube cuddle get nodes, 2, 9 f57! This is our gateway, so our gateway is really the host, because you can kind of think of like a cube node as like its own little kind of sub network in a way. So we'll do that that worked!

A

Okay, let's see if my my theory is working or if I'm just atrociously off here and it's not going to work.

A

Okay, waiting for service to become ready. Can I curl that service? Let's see here.

A

Nope not a happy camper, 1065 yeah, that's what I'm wondering evan, if the lo, if it requires a load balancer, um because if so, I'm not gonna have a load balancer, um but if it's hard coded to check for an lb ip, I was hoping it would just look up the the service ip perhaps, but because I mean in an on-prem environment like this, I might not actually have a load. Balancer cube cuddle get service lb, um but if we do check this out, what is the what's the default port 9701 here?

A

Let's, let's check something out real quick, um so boom 9701 nope! No luck, yeah! Okay! Well, we might not! We might not be able to get to the ui then, because I'm not going to have a load balancer, that's going to provision in this environment, at least unless I install metal lb, which could be an option yeah. I guess it must be exclusively looking up the external ip okay, we'll see if we can figure that out and work around it um as funny as it is, could you just slam metal? I could turn metal lbn.

A

I guess um really. I can't uh so. The cli will need interesting, huh I'm trying to think like okay, okay, yeah, I can deploy metal lb. We can see if that will work, um all right so I'll, be.

A

Deployment, okay, cli can launch a local server yeah. I don't is and just to check on the cli. There's no easy way for me to say: don't rely on the external ip. I can't just give it because in theory it can totally talk to this. It's in my routing table, but I'm guessing it's maybe hard-coded, to look up external ip. Is that fair?

A

All right? Let's see if we can deploy metal lb um yeah, it's waiting on the kate's api. Okay got it got it okay to save the api for cube, it wants one, but you could run the server locally on docker, though yeah. I don't really want to spin up kind.

A

To be honest with you, I don't want to install it and set everything up, I'm a little adverse to metal lb as well right now, but we're going to give it a shot, see if we can get it uh see if we can get it bootstrap, so installation, um you know if it can grab a vip and respond to the arp.

A

We should, I think, we'll be looking pretty good, so you can achieve this by editing cube proxy in the cluster, if you're using q proxy in ipvs mode, I'm not doing that and set this nope nope install metal lb daemon set all right. Let's, let's see, if we're going to have any luck here, it probably needs an ip range somewhere, though right um waypoint install yeah, I mean I could install it on docker too. I suppose I'm kind of stoked to put it on kubernetes, to be honest with you, though, so we'll see.

A

If we can get that, if we can get that working, if not, we can uh we can pivot and go into docker no big deal um cool, all right, metal, lb controller deployment, speaker, okay, all right I'll! Tell you what here's here's, what I'm gonna do! I have. I have a running theory here, um so I've got I've got the server in there and what I want to see real, quick.

A

It says to initialize waypoint. You must run the init command, so I just want to I kind of want to understand real quick where it's accessing that thing. So um I'm going to humor me real, quick, I'm going to try something out so for the node.js app that was in the examples directory. It's got some. It's got some bits here, so I just want to see it kind of fail from the deploy perspective and then maybe I can kind of reverse that back over to metal lb.

A

So if we do, um if we cd into waypoint examples and then we go into kubernetes um gabrielle, you said after you install. Oh, you create the config map. That's got the ip range in it, yeah, okay, cool, so we'll, probably just throw metal lb in and we'll put that config map in. But I do just want to check out this app real, quick and once I see it blow up on this level, I promise you I'll go in and try to make my service routable no big deal.

A

So um if we do a quick look at the hcl for waypoint all right, we've got the node.js app here and it looks like these are the three things that we were talking about right. There's the build step which we can point at docker, the image step, the tag, step or not step, but the image specification, the tag and local, and then we have got the deploy which is using kubernetes. So that makes sense and then the release itself so we'll have to dig a little bit into these.

A

But what I want to do is I kind of want to run the build and see exactly how it blows up and then we can go back to uh to metal lb. So, let's, let's start off here, um it looks like we're going to be using build packs by default. So I'm guessing there's no docker file in here yep looks good and in the commands I think it said we could run a waypoint init.

A

So let's go ahead and try that out real quick, so we'll do a waypoint, init, okay and now we can do waypoint up which again should blow up. It looks like waypoint and net correct me if I'm wrong made this datadb file and that might be about it. We've got the waypoint hcl right here, which we just looked at now. Actually, now that I think about it granted, I don't even think we'll get to this step.

A

I think it'll probably blow up in the pack stage, because I won't be able to pull from a local repo, so I'm going to put in I'm going to put in my fully qualified docker hub url. Hopefully that works and we'll call this tgik, and we will apply that or up up that so waypoint up right.

A

Okay, so building the image um docker's not running on my system, we'll solve that docker systemctl, start docker looks good, okay, cool cool and let's go ahead and try this one more time. Waypoint up so build should work. I think deploy should fail right. Does that sound about right people, uh given that it can't access that server?

A

Okay, waypoint installer message says you can set the server address manually, maybe try setting it to a pod ip that works, ah noelle good call! If that's the case, I might try.

A

Oh okay, I see what you're saying evan, that's using a local server got it cool cool. All right, hashicorp provide their own, build packs, um looks like evan, says they're using the cnb. uh The cloud native build packs for right now, um ymo, you probably saw the change already the the chat might be delayed, but um I uh I just pushed to docker hub basically, so I'm assuming that that fully qualified thing will make the same assumption that uh build packs normally would which is that it wants to push to dockerhub now. This is interesting.

A

Look where it's connected, so is it talking to the server to deploy the deployment? Clearly, not right so for for the cube deployment piece. Is it talking to the cube api server because that's what it looked like it was. uh It was talking to there yeah. Am I wrong?

A

Let's see so cube cuddle get pods okay, so it's hitting the kate's api. So tell us evan. If you don't mind, sorry that I'm like totally quizzing you here, what are what are the things the server is going to provide? So it sounds like the dashboard for sure. Is it doing stuff to handle like the rollout components as well?

A

Maybe that's where the extra functionality is handles. The web ui okay cool cool, and I definitely want to definitely want to try that out stores all the data. Okay, interesting about the artifacts. Okay, okay, got it and then coordinates jobs, so we could say okay, so the the deploy.

A

Oh there's, a runner model. Okay, I gotcha all right, fair enough. Okay, so we'll have to get that that server running and see if we can, uh if we can get that wired up, but it looks like it looks like uh it did, deploy the app okay, um it looks like we've got the the node.js app. If we do a quick, um let's uh get the well actually we'll check out the uh the docs real, quick waypoint up, deploying configuring. It gave us a deployment. Oh the deployment.

A

Url came in my terminal, didn't it so this is my deployment url. That makes sense. Okay, um what you are using now is just a server that is run by the cli whenever you run a command. Okay, I see um and then code. You said the server also provides external accessibility.

A

um If I deployed against the server, you could share a url here that all of us could hit interesting. huh Okay, we'll just see if we can get that server up or routable, so that I can so. I can hopefully try that out. Okay, so we've got the release url. Let's just make sure that the app I mean the app seems to be healthy and fine, but we'll just make sure that it's that it's good. So if we do cube cuddle, get pods uh we'll do o wide.

A

This one got dropped on four s7 s, cube cuddle, get nodes o wide, and this is force 4s7s.

A

So here's my ip address all right, good good and we're just going to do another one of those hacks real, quick, so we'll uh make that routable so route that here's my ip this one I could probably just use cube proxy for frankly, but that's no fun. Let's grab that! Let's get this um 92 is the ip here so boom and then we'll do this.

A

Okay, let's see here.

A

What am I doing, 92 is what I wanted cube cuddle get nodes o wide. I think I'm getting my copying and pasting mixed up there. It is okay, good deal good deal, we're good. Okay, let's see if we can reach this url, real, quick, okay, so all right good deal. So we got you know we got. We got a little ways.

A

um We still got to figure out the server thing, but it looks like we were able to go in wire up waypoint to some extent um and then be able to just deploy the app and and roll through it. Just like that, um let's see what else you all got going on ymo says is waypoint, something that you can start from a ci cd pipeline and cody says yap and I think cody. I saw it on twitter, but I forgot to add it to the notes.

A

Somebody from gitlab has a blog post on it too on how to kind of integrate waypoint with ci cd. So if anyone has that link and wants to throw it in the show notes feel free to do that.

A

While lead said, you can try to roll out a new release, yeah good point- let's see here so it says: try making a change of text locally and then doing waypoint up again.

A

I'm guessing it'd probably be a lot cooler if we can go in the server and see it too, but let's just let's just validate that real, quick and then we'll get the server thing working. It can't be that hard. So, um let's see am I still in the node.js app yeah. It looks like I am so if I go into what's the try making a file, try making a change this text locally?

A

Okay, where is the? uh Where is the the text coming from? Is it in public nope? Is it in views pages partials? Maybe let's see header, I don't know anything about javascript, okay pages index. Is this: it hey, okay, cool all right, so tgik waypoint is awesome.

A

Let's try it out. So we'll do that change and then you know again keep in mind. We got kind of a whole pipeline that um waypoint is sort of. I guess orchestrating. You could say so: let's go in and let's do a waypoint up again.

A

Okay building project was not found. Oh I'm in views now duh all right, so we'll do waypoint up it'll rebuild and then I'm guessing. If we do a watch for the pods, so we will watch cube, cuddle pods get pods.

A

All right, I'm guessing once we get through the build phase, we should roll through the update.

A

All right so keep an eye on that for a second and then see if we can, uh if we can get this roll. Oh there we go cool coming right through pulling up good deal.

A

A

A

A

All right cool stuff- let's see if we can get this this server, set up and good to go, we'll give it a shot with uh with metal lb and, if not worst case scenario, we'll do it in my local docker damon how's that sound, so let's go in and get metal lb deployed. So where did I put that deployment? Okay, customize.

A

Trying to find that supposed config map, I need to worry about so cube. Cuddle, apply, namespace, metal, lb secret, key okay looks good.

A

Let's try this lb system memberless secret, key okay, the installation manifest, does not include the configuration file. Okay, that's probably what I'm looking for. So I just need to go in and probably put a range in is what I'm thinking? Okay, more so you said, metal lb needs another episode for itself. I totally agree I feel like we I'm surprised we haven't done metal lb. Have we not done a tgik on metal lb?

A

I'm surprised, okay. So let's do this thing we'll go in and we will try to get metal lb bootstrapped so get out of waypoint real quick and we will cube cuddle. Actually, let me just do the metal.shell here such a metal, shell script. Okay, so we will add that as executable and we'll run metal, shell, okay, good deal.

A

Let's get this thing. Bootstrapped cubecuddle get pods inside of namespace metal. What namespace name did it put get namespace metal lb system, okay, metal, lb system, sweet?

A

Okay, we've got the speakers up, which should be able to respond to the arp request. um Everything's running the heck is my config map at so cube, cuddle gets. Did it deplete configmap by default?

A

No, it didn't okay, so inside of their docs here, there's got to be a config map. Okay, so you can saw metal lb with customize. If you want to use the config map generator.

A

Let's see here, configmap is named config.

A

Do you mean that it's already in the cluster, or am I just missing it somewhere? Let's see.

A

Joe says no metal will be for metal anything yeah, the virtualization company isn't doing enough metal related stuff is my take on it. um You got to create it where the heck is there an example of the thing you need to define the config map give me ah here. We go hey. Okay. Now, if I just choose an ip, that's actually routable on my network, I think we might be in a good spot. Let's see if see, if I can actually do that. Alright configmap.

A

So this will be metal, lb, configmap.yaml, okay,.

A

Good deal all right, so what is my range? That's safe. So in my network, 16, okay, I should be pretty good. So I'll, like choose 168.5 or something like that. So 168.5, we'll just we'll just cross our fingers and hope no one's using this ip uh again, don't tell anybody, 100, 168, 5 200.. What do you all think that seems like a legit, uh legit range metal, lb system, cube cuddle, apply, f, metal, lb, okay, all right now,.

A

Hey I have an external ip: is it routable? Is it actually routable next question? Okay, so let's go here, um would the? U would I be able to? Would I be able to just hit this server? I don't know if the server like has a ui by default or if I have to do something. Thank you, metal lb. I agree. Thank you so much and as as rich says, nobody is allowed to tell my boss that I am messing with routing tables and using metal lb.

A

Okay, uh it has one by default. uh Oh is it does it need the host name? I doubt it um well, let's see real quick if I do metal lb okay here, let's, let's back up a little bit, because I've kind of mixed things up now I did a waypoint install before which I'm guessing waypoint install is going to fail. Now right, it probably will see the stateful set and blow up creating kubernetes resources, blah blah blah. Oh the ui is 972..

A

Thank you cody. You may have just saved me six hours of figuring out why I can't access it. Nine- and I probably should have known from the service right- nine seven zero two is that right, hey okay, client send an http request to an https server. That seems problematic.

A

Hey, I don't know hashicorps. I trust you what's going on with this cert. Okay, oh! No! What happened? What I thought, but did I say, go back. I need to look at what I click on advanced is what I meant to say, accept and continue check it out. Everyone pretty cool, all right to get started. Please authenticate um run this to generate the cli pretty cool um here we go. We've got a token.

A

All right authenticate huh paste token authenticate what you, what are you doing? Waypoint? I think it's uh I'm doing something. Weird, um that's not going to work, because your contacts for the cli was okay, good point. So here's my proposition oven, let's, let's delete waypoint and do install again. What do you say? I feel like just to prevent me from screwing up more and more as we keep going um so get pod all.

A

uh What do we got going on here? Let's do uh is, is uh waypoint devs is the easiest way just to delete the stateful set, or is there like a more robust way to actually do a cleanup here? Oh there is a cleanup command. Sorry, my chat was screwed up there for a second, um oh delete. The staple set. Okay, lovely sorry reading chat wrong here. Let's do it so we will cube cuddle, delete our stateful set for waypoint server.

A

huh Isn't ss a shortcut for stateful set. Am I losing my mind? I thought that used to work.

A

huh Oh, it's sts, ah okay,.

A

Not found oh duh, because I have the id of the thing on the end there. Okay, that was deleted um and you said the server service. Should I delete the service too cube cuddle get service.

A

We will delete the service which is waypoint as well. How's that sound getservice seem good. We've got the service deleted. We've got the server deleted. Okay rex says never use aliases. You've called me out. I am a hypocrite to be fair. Technically, it's like a shortcut right, but no, I hear you you're not wrong. Okay, so let's do the install again um ah delete the pvc too. Okay, we need a cleanup command. I think we need to clean up command.

A

Okay, cube cuddle, get pvc good good, call, keep cuddle, delete, delete pvc data, waypoint server, all right lovely here we go fingers crossed everyone. Waypoint install I'm trying to remember all the commands here. Platform equals kubernetes, except tos, hey, okay, cool stateful set is reportedly running. It found the pv, while it's doing that we'll do a cube. Cuddle get service, an external ip did get provisioned, so hopefully, hopefully it gets past. Whatever it's doing here. um Let's just double check, so we've got 100 here uh we'll go here, we'll do 9702 again wait.

A

What was the ip nine? Seven? No, nine, seven zero, two yeah okay! Maybe the server is just not up yet there. It is okay. Now it's up and hopefully my little command line friend you find my server.

A

Okay, waiting for service to become ready.

A

Well, you said a knit from the example dirt. Do you mean for the app because I think that's already uh yeah cody I did but this time it should be good because I had the pv still so oh wait, you might be right, you actually might be right. The pv says status failed, cube, cuddle, get pvc ah good call good, call that maybe that's what it is. um Hopefully I can beat the timeout here. Let's do this uh example pv number two uh and we'll apply that pv.

A

It's funny that the stateful set started, though, why would the stateful set start if the pvc was failed? That seems like a weird kubernetes thing: uh okay, 222 is created, yeah yeah eric my uh my can is green, which is which is pretty rad right. That's pretty cool, um okay, so hey whew! That was it. Why did the stateful set start if the pvc was failed? Am I losing my mind everyone? It should fail right if the pvc didn't bind.

A

I don't know I must. I must be losing my mind here: anywho, that's, that's a kubernetes problem. That's not a waypoint problem! um Nonetheless, we've maybe got an uh ui that we can use now. So let's check this out pretty cool stuff, we'll authenticate we'll do the waypoint token we'll hit that and let's grab the lovely token here, incomplete response all right josh.

A

What did you do now? Okay copy. That waypoint token.

A

Let me just let me just pipe it into my clipboard to make sure I'm not uh doing something silly here: okay, okay, we're good! So there's that oh there's a command for waypoint uuid yeah joe what's funny is I have a csi running and I don't have a storage class configured. So uh I don't really know I I don't know what I did. It's probably something on my end: dang it incomplete response. Okay cody says waypoint ui authenticate. Should I try that um waypoint.

A

Waypoint ui authenticate, okay, something popped up in a browser. You all can't see. Let me see if I can bring it over secure connection, failed.

A

um Weird, let me bring this url over here. Let's see here, okay advertise address web ui address, cert stuff, probably assert stuff, I'm living my best life over here. um So this is what uh this is. What waypoint ui authenticate does. Am I losing my losing my mind again for the second time today, security error, reissued issuer serial insert without the same import. Serial cert is there? Is there any chance there's something wrong with the search on yeah? I wonder if it's an ssl or two evan.

A

Oh, I need to bypass assert, learn more: oh boy, mozilla. Why are you making my life so hard? Can I just can I just bypass it is: does chrome make this easier to bypass by any chance? Let's see.

A

I love you firefox, but sometimes I hate you um here. We go hey.

A

Okay, here we go when in doubt just sell some data to google and uh call today uh 100 hey here we are okay, so we were having an ssl error, you're right, um okay, so now we probably can just roll through this example again and try to bring all this full circle and we might be able to wrap up on that. So projects are looking good um and nuno. You said you have accepted the previous cert. Oh, the new has the same cereal. That would make a lot of sense. Then.

A

So it's freaked out because before I got to it, um maybe maybe that's what it was doing. Okay and rich said he's snitching about bypassing certs that'll that'll, be my uh that'll, be the end of me there about all the bad things I did accepting untrusted cert, you know um cool all right. So, let's deploy the app again and let's see if we can get this thing all wired up. Just like yusuf said yep: let's do it. Okay, so we'll go back here. I think I still remember the commands.

A

It should be pretty simple, so we'll go in to the waypoint examples.

A

Was it was kubernetes node.js? It was waypoint h, not waypoint. um Oh, I already have it wired up, because I did that before right. So as a reminder, here's what we've got going on um we've got this uh example. Node.Js we've got pack build, we've got the deploy. You know it'd be interesting. I don't know if you got if you all have an example on the docs already, but is there a way in like the release stage for me to add a load balancer to the uh to the deployment too?

A

That could be kind of cool now that now that you've forced my hand and made me deploy metal lb, I might as well make good use of it. Okay. So let's do waypoint up waypoint up uh node.js was not found waypoint in it was run with this project.

A

um Oh evan says the default load, balancer, okay, cool, so init. Oh, I was using a different server before duh duh duh. You all are right: yep, waypoint, init project's initialized, again waypoint up all right moment of truth.

A

You're, probably all thinking josh. If you would have just freaking made the ui work, the first time we wouldn't be doing this again, but we learned some cool stuff. You know here and there file it in the category of useless information. um Cool, so build packs are going.

A

I'm deploying up to my docker hub account, and now we're deploying should I be looking in the ui. Is that the best uh best place to look ah check it out? Example: node.js ah all right, let's see what we got going on here, so, okay, so tell me if I'm totally off here this ui long term is this. Is this a ui? That's going to be something the developers will probably be using um I'm guessing. So I'm guessing it's not like an operator only type ui, um the terminal screen.

A

Do you mean the uh the like loading bits that it does while because that is pretty freaking cool? I, like I like how it's like, pending constantly. um Oh, it looks like we got some okay. We got some more interesting stuff here too. So we've got uh the release url, which is from before, and there's that deployment url.

A

Can I deploy app from ui good question? Flash, let's see if we can make, can we do it from the ui. It looks like from what I'm seeing right now, maybe not maybe this is just reflective of that. Oh check it out. It's pulling the app logs off um okay, evan says the ui is meant to be for both operators and devs, because when I was coming in here and seeing it, it's like okay, my builds are going to be in here. Oh, this is cool.

A

I love how it has a little build pack logo and the kubernetes logo. I get really impressed by the things that are not really substantial, but seeing these little logos here, it kind of just tells a story right because you can have like we were talking about really early on those different. um Those different back ends right and kind of see how you've how you've plugged into those from this ui- um but I guess: huh okay, okay, okay, the the dots are slowly connecting here.

A

If I was running an ecs, the ui would be a totally separate server right. So is there's no like federation today, I'm assuming where I can see universally, where waypoint has put all my stuff like, I would think for each provider, I'm gonna have a unique ui. Tell me tell me if I'm totally wrong there I could be. I could be off.

A

um So this is cool built from pack kubernetes. Oh evan says you could use one server, okay interesting. So this server, that's running in my cube cluster. Are you saying that it could plug into like an ecs back end? To do the deployments? Is that fair?

A

All right, um so we've got logs, that's pretty cool. um We've got exact, oh it! Oh! It's an! I see it's an abstraction on like docker, exec or cube cuddle exec. Let's check that out real quick in the in the command line. So if we do waypoint, okay, I'm guessing maybe it'll just know because of where my app is waypoint, exec.

A

Okay, I'm connected to deployment v1. What does that mean? um Let's see here connected, do I need to tell it the command. I probably need to tell it what I actually want to run there. We go sweet. Okay.

A

This is really cool so by, if you think about this for a moment, think about how neat this is from a developer's perspective like I don't need to worry about cube cuddle.

A

What's the name of my pod or cube cuddle, blah blah blah where the heck is this thing running, based on the context that waypoint has from this local directory, where I did the up just running exact is: is that's all abstracted in short, like it's just you exact, and then it wires it up and I'm guessing hypothetically, whether we're doing a docker, exact or a cube, cuddle, exact or whatever, like that's now, not a concern for me as an end user.

A

I just run exact and I just got the thing and I suppose the same principle is true for logs. Like think about this, don't need to find the pod name, don't need to figure out what namespaces it is in blah blah blah. I just run waypoint logs and there are my logs. That's a that's, a pretty uh pretty sweet experience from a developer standpoint right, yeah! The I agree with that exact is uh that's slick. I'm into that it'll.

A

Be cool when it shows up in the ui too, um all right, so if apps reach each other sure evan says the biggie is at the end the entry point to work. Okay, so the architecture thing about the entry point, the piece I didn't really grok is probably where the keys are there, um let's see here or john chase says: can you exact in the ecs ones? That's a good question. I am not sure.

A

Okay earlier, when you asked what the big thing is, it's the concept, a consistent weight, exactly just like cody, said that I think is perfect. It's like you can have all these providers right and you can have um an abstraction that takes those takes needing to learn those details away from developers, and I again like I'm trying to be somewhat straight here. There's there's trade-offs to that. I think I think in both directions, but a lot of people are trying to bring experiences like this to people using their cube clusters.

A

um I remember actually, I remember an example. This kind of makes me think of you all might have heard me talk about this. There was a kubecon video and uh if anyone can find the link feel free to put it in the show notes, there's a cubecon video from who I think was uber, um and they were talking about their abstraction system that I think they called something like pelletin and this thing abstracted mesosphere and kubernetes. It was really a quite similar concept like in-house.

A

They had to build this developer experience that abstracted these two underlying providers. So not only is it something where, like in theory, you don't have to become a cube expert. It's going to give you flexibility as an abstraction typically does for there to be options in the underlying providers for there to be a change in the underlying provider over time where the people plugging into your interface, don't need to be worried that you're changing things under the hood.

A

In theory, it's the devil's in the details there, but, like you kind of see where I'm going, it's kind of a pretty compelling stance on how we can build systems and how we can create developer experiences on top of them. So this is uh this. Is some interesting stuff? Okay, so what else should we look at here? Is there anything? That's, uh oh sorry, wrong, wrong browser waypoint, so we've seen exact um releases, I'm guessing. This is just a hash.

A

um Here's, our here's, our ip address, we've already gone to the website, so we've seen that kind of roll out. I don't know what else do you all think? Is there something else we should? We should take a look at here. We've got logs, we've got builds. Let's go back to their tutorial and see if there's anything else that they uh that they called out in here. I'm impressed, though I think this is really really cool. um Tell me tell us about the release url thing, because I saw that show up in my terminal.

A

I think the release url, where was my release- url okay, so mainly curious, flamingo, v1 waypoint- is that a real dns record? Is that a real thing um trying to figure out where that's coming from so.

A

It says paste it into chat and tell oh, you all can hit it. So that's a that's! A url just put it in chat. That's a url y'all can hit.

A

Oh cool, interesting, stuff, okay, yeah and I'm guessing that behavior is obviously just a default that we can turn off but interesting, stuff. Okay, so you think about like url sharing and dev environments and stuff like that right. That's that's! Pretty interesting! huh Yeah! Okay!

A

I dig it. I wouldn't keep it on in production, but I do dig it it's it's a cool feature set yeah, wow yeah! That's uh I don't know um I don't know uh cody or um or evan, if you, if you all, have like an architecture doc for like how that works. Maybe I missed it in the docs, but that would be really cool to to read about how it's like broadcasting that, because this app is certainly running in my private network, so I know y'all aren't ingressing to my home lab right now.

A

I can tell you that without question, so I'm guessing the server is doing some type of proxy. Is that oh some says check logs? Let's see, uh check logs.

A

Let's see here now that I got two browsers, I'm all confused. Oh now, the logs don't say anything else other than listening. um The entry point connects to the cluster okay yeah. I gotta read about that. That entry point thing which says: I'm in your lab, you people in my lab. You be careful uh next thing. I know I'm gonna hear the fans on my servers just like ramp up and uh you all will become rich with some bitcoin. I think not that I have enough hardware to actually harvest anything meaningful.

A

All your labs belong to us yeah. That makes sense evan, so you're forwarding the connections back to the internet. Okay got it, got it yep yeah! That's that's a pretty cool idea like you, you think about. I mean again I'm not trying to like oversell anything, because I mean I don't I don't work at hashicorp. I don't care, but um you just think about how much uh rigmarole, how much stuff we go through to like share a url with developers and um well.

A

You gotta be careful with how you like allow this feature to happen in most more serious contacts. That's pretty cool, like hey. I just pushed this code up click. This link boom they've got it um and you know there might be some work around like uh providing internal services that do this and stuff too.

A

I don't know if you all have thought about that, if it's only something that can hook into like the public internet or if it's something that like we could run a little internal server in a network and then have the same experience for inside of the uh inside of the intranet. You know what I mean so uh alex said to the electric electrician, install a 20 amp plug for your servers, um yeah yeah I've I've got some.

A

I've got some backups, so uh hopefully the power won't give out for the next 20 minutes, while we're on tgik still, um okay, that is pretty freaking cool. I'm impressed this is. This is really neat um okay. So let's go ahead and check uh choco asked uh if oh evan says it's all oss, so I'm guessing, we probably could do that internal use case pretty easily um okay, so we've got all that in place.

A

Okay, maybe maybe, as a last step and I'm open to ideas here, if you all have have more, but maybe we could try, um I'm trying to think how this would work. Could I hook into.

A

With the same server, can I hook into my local docker daemon, I'm trying to understand I'm a little confused. I think I'm I'm misunderstanding. The entry point thing I'm trying to think how like, if I added okay, let me let me make this more concrete, because I'm just I'm just blabbing at this point. If I did uh waypoint.hcl.

A

Can I can I do something like this? Is this a thing where I can have it deploy to two providers like this.

A

I mean it's kind of an obscure use case. I know, but I'm just trying to think if it's pointed at the server that is on my kubernetes cluster.

A

How would that work exactly? Oh? Well, it's gonna! Okay, okay, I think. Oh, it's not a thing right now: okay, evan you're, stopping me from making a mistake. So it's not a thing right now, but I'm guessing. I could switch it out to run docker in the case that you can't have two there. Okay got it well, let's, let's, let's experiment with something real, quick just for the hell of it.

A

So if I do this- and I probably should be reading the docs to see if there's different fields here for docker um release- use docker- I guess um so, let's see if we can, let's just see what this does. It probably doesn't make any sense. So let's do waypoint init um failed to validate plugins required plugins, oh ducks. I didn't run server against dash dash plugin, so yeah. That's interesting, hmm interesting, interesting, stuff, an unsupported argument! Validation!

A

huh Oh wait! Is it the unsupported argument that it blew up on the this validation check? Ensures you have all the required plugins available in the configuration for each plugin if it exists, it's valid, um oh! It's because I had probe path. Okay, that makes sense so waypoint, let's get rid of probe path because that's not a thing in docker per se.

A

All right does this work now, oh wow, okay! So now, if I do deploy way point or not uh deploy up sorry, I'm still in my I'm still in my kubernetes mindset. um Okay, so I have this I'm guessing. This is my local docker daemon! That's the best thing. I could think so. We'll do a quick watch here. So in the bottom is my docker daemon um we'll see if it oh, hey there, it is pretty cool okay, so you might have noticed. I use the command waypoint up.

A

That's what's cool about this right. um It went to cube. It went to docker.

A

Theoretically, we can code providers to probably do very, very specific things here, based on how we want to deploy and what we want the behavior to be so. I've got that wired up now: here's, okay, here's where I'm really interested ah check it out. Okay, so I see okay, I think I'm starting to grok the relationship between the server component. Now it's starting to make a little bit more sense. To me, um I'm not going to say I understand how all this works yet, but I think I'm starting to kind of understand the mapping here.

A

So I still have the server running in my cube cluster, as you all probably know, but how it's kind of going through and interacting with docker and keeping the server up to date in a way is still can totally still run inside of my my kubernetes cluster, that's not a problem at all, so I can see my docker deploy. I can see my cube, deploy no problem all right. This is pretty pretty cool and then exact and all that stuff should work the same just like we talked about.

A

If I come in here and do waypoint exec someday I'll remember, I need a command, but maybe I'm using docker exec. Maybe I'm using cube cuddle exec, it's abstracted, there's a plugin that defines how to satisfy this api. This cube, cuddle, exec or sorry. This waypoint exact api, which then can enable me to not have to worry about it.

A

Pretty sweet um gabe, you gabriel, you said hello, go edit the app and deploy a v2 josh. Okay, fair enough. I hear you let's, let's deploy, let's deploy a v2 real quick, so um we did we did this earlier, but we didn't get to see it in the ui. So, let's, let's check it out, we'll go to waypoint, um I'm going to go back to kubernetes for this one. um So this is kubernetes.

A

um This release will be kubernetes. Hopefully it's okay, that I don't put a path in here and oh, it's v2 now good point right: it doesn't. I guess I'm getting my providers mixed up. It already should be v2 right. So if we go to releases there is v2 yeah. This is it so if we go into v1 um I'm guessing at some point, this will probably give us some more details about this, but we had v1, which was replaced on cube.

A

We've got v2, which was deployed on docker and even if we made a code change or whatever it might be, I'm guessing it's. It's sort of the same, the same mechanism here which is pretty sweet and then, if we did all this with a docker file rather than build packs, it would do that as well. So the yeah exactly walid v1 is cube. V2 is docker, that's it okay, oh evan did call something out that I didn't think about so waypoint up.

A

If I'm hearing you right, evan is doing the whole stack, it's doing um build, deploy rollout um and I could have just done so. If we like do this real, quick, we'll do uh views partials, header of geez. You just did this a second ago, josh partials. No, it was pages right there. It is okay and then we say.

A

It freaking works, then in theory. Well, no I'd have to build again duh because it's the pages the page is compiled into the image. What am I thinking? Okay, that doesn't make any sense, uh waypoint well we'll we'll deploy this one anyway, so waypoint, um and if we just do the stages one by one. So if we do waypoint, is it build yep, so build we'll, build it back up, I'm guessing it's gonna package up, all the all of the uh javascript and html assets.

A

Here, oh and I'm still on my uh I'm still on my local, my local docker endpoint. But what we'll do is we'll change it back to cube now and we'll go ahead and uh deploy it to cube, then so cool cool and that by the way that docker thing that just popped up there that I think that was just the build pack thing. I don't think that was uh necessarily waypoint okay. So now, if we go back to the hcl, which was kubernetes right, did I switch that already uh build yeah? Okay?

A

So now we do waypoint deploy all right and then we go back to the ui, and I think in a moment here we should see the uh the v3. I guess in this case come back, so pretty cool all right and this gabriel. This hits your your use case of updating something because I actually updated something this time. There's v3 right in kubernetes.

A

um Oh, it takes us right to the app and check it out. Waypoint is awesome, pretty cool all right. What else are you all talking about? um Oh matty, you asked if there's something similar to a terraform registry for plug-ins. That would be pretty cool and interesting, especially as we build these out. What's gonna be super interesting about how this plays out.

A

I mean I'm sure, there's like a canonical set of um I'm sure, there's like a canonical set of functionality that we probably all want, but I would think in organizations a lot of a lot of orgs if they were to truly use this, at least- maybe maybe I'm maybe I'm biased, working mostly with like telco and financial services, where things get like really specific.

A

I feel like a lot of people, will either extend or write their own plugins to make a lot of this stuff happen, because there's going to be so much regulation and preference and things they have to get through and also just like how you yeah it's super interesting when you think about it right, like the pattern of how you want to deploy the thing so like a lot of folks will have opinions like oh, I use this ingress controller and I need for a roll out for it to be this very specific, ingress definition where it looks up this value from this place and blah blah blah right or or does you know, does x, y and z very specific to how the cube stack is under the hood.

A

And ideally you don't want to expose that to developers, but I would think with whatever model waypoint's using you know you could go in and you could build that functionality out and that behavior out and again it's all about providing this experience and- and just you know, working on the back end to to get some of those intricate intricacies kind of pulled together. um Oh but josh, you said check the plug-in section in the website. They are cool with grpc implementations cool.

A

Let me let me look at that real, quick and then we'll probably just wrap up roughly here today.

A

Well, oh, while you said it was the v, is it the v2 message or is it v3 couldn't find waypoint url? Oh maybe I did I have. Was it just delayed or something is that what happened here? Let me go back uh yeah. I guess I did say something different for v3 right, huh tgik waypoint. I wonder what I did there.

A

It's probably just cached wally, you said you saw it update. Is that correct? Let's just hack, let's just open a fourth browser right. Why not? um Here we go? No! It's still there. huh I thought I'd change that anyways okay. So let's check out the uh the plugins the plugins here. So where are they? Are they in the docs for waypoint?

A

uh Oh there they are plugins using an external plugin.

A

Authoring a plugin, so I don't know if there's a list somewhere with what exists or maybe the plugins are just. Oh here's, the built-in plugins, my bad, my bad here we go um plugins, okay, cool, so here's the plugins you've got and then yeah I mean it's it'll, be so it'll, be so cool to see how hashicorp kind of plays this out, um because it's like you think about the base level functionality for cube and people will probably want that.

A

They'll want to be able to play around with waypoint make it work on vanilla, cube, make it work on vanilla, eks and then, when you start doing like the real stuff, the stuff that's going to be like in prod it kind of shouldn't blow up or have a terrible ux, and it probably will have ultra specific needs. Then you just either extend the cube one or you build your own right.

A

So this is, this is pretty cool time will tell how many how many people are starting to use waypoint for for all the things right um and joshua. You said uh waypoint.com docs for extending waypoint is that is that not where we're at with plugins using an external plug-in authoring a plug-in? Oh, is at the bottom here extending waypoint.

A

I see okay, cool cool, so it's just a matter of probably satisfying. Some grpc calls pretty sweet.

A

Okay and flash, you said: can I use uh waypoint to run some ci cd and yeah? The answer is yes, I'm not an expert on this, but I saw a blog post go up from someone in gitlab um check that out. If you get a chance it was somewhere in twitter. I gotta find it. Someone might add it to the show notes too, but yeah you can there's definitely some way you can do waypoint integration with ci cd, um yeah yeah, the netlify plug-in that would have been.

A

That would be an interesting one for us to try out actually cool uh on the left side of the page scroll down to automating execution, okay, automating execution, oh cool! There's your answer: good catch, so gitlab, ci cd, all that good stuff. What's interesting about this too, and I I I really dig this like this approach and what it's doing, I think there's been um and tell me if you all agree in chat, and if you don't I'm, you know, don't be honest. I I can take it.

A

There's been a big push for like git, ops, right and I feel like in a way when we talked about this principle of abstraction right and like what's the right abstraction for your organization. I think, what's funny about. Gitops is a lot of that kind of reconciliation approach based on a git repository backing. It has been a bit of an abstraction for a lot of folks. You know um it's kind of like uh it still requires you oftentimes to write a yaml file and check it in and all that stuff.

A

But in theory, if you do it fancy enough, you can find ways to pull stuff in and- uh and you know, just kind of make it run in the cluster um which- which I think a lot of folks- are kind of using as an abstraction bit so it'll be. What I'm getting at here is what's interesting is like. Should the get ops model gain steam?

A

I wonder um if it will be a competitor to abstractions like this and not just waypoint but abstractions. That kind of like have this kind of dev experience, that's similar to it or, if there'll be like an integration point that makes sense like perhaps the idea of your deploy plug-in and now I'm just shooting from the hip. Here, I'm not saying.

A

I think this is a good idea, but your deploy plug-in generating the static assets you need to then check in which then gitops takes over from there, the notion being like reducing some complexity and how, whatever engine you use for git ops, wires. Everything up still letting waypoint provide that experience, but the target being a git repo for the end result manifest, which then gets picked up and managed from some type of intra-cluster agent.

A

um So yeah, I don't know- and uh it's just kind of interesting how I I guess, what I'm trying to say in too many words frankly is like that one of the next frontiers of kubernetes is like finding how to abstract it.

A

Well, I really, I really think that's the case and I'm not just trying to be cliche and say, like you know, oh you know it's it's about what goes on top of kubernetes, which is totally valid, but more so that just like now's the time for us to find like what are the right abstraction options, and I think there can be a lot of them, and this is a remarkably compelling one. I think, as far as the experience that we've seen so far and we're just in early days, it just seems like this.

A

Get ops and other stuff will be, um will be interesting to see how this plays out. um We had orchestrator wars. Maybe we'll eventually have uh orchestrator abstraction wars, we'll see uh cool so walid. You said: how would it work if you have different clusters for dev verse production?

A

That's yeah! That's a good question I'll! Let cody cody throw some throw some details in there. um Okay and ivan. You said you wonder about oh hey ivan by the way, I wonder about waypoint interaction with cube, manifest rejected by admission, exactly yeah.

A

So it's it's that's that's kind of what I'm thinking ivan, like you think, and I know you work with like some some companies and stuff, so you've probably seen like how, when cube clusters, go from just vanilla right to becoming like way more intricate, like there's a mission, control in place and psp's in place and resource requests, quotas and all these things- and you know I I think- and I could be wrong- I'm new to this stuff, but like I think where some of that value add is like as a platform team, if you're gonna adopt waypoint, maybe you really think about how you're going to build a plug-in that fits into that model?

A

And I know it's not your exact question like how it interacts and how it handles rejection is a whole other story, but, like I think you can really build something. That's compliant because as a developer, it sucks when you deploy something and admission control, blows it up, and then you have to spend eight hours learning how kubernetes works to understand what the hell uh security contacts, user id1001 is wrong.

A

uh Error means you know like we might know and love that kind of thing, but a lot of people don't so you know if you have this abstraction with something like waypoint and you can have easier paths to do. The right thing you might be able to.

A

uh You know again provide that better experience and not have to make it such like a defensive stance where it's like um you know a lot of people run into it and it kind of blows up and it's being from personal experience a lot of times again when I help especially financial services groups. The protection that we have to implement is crazy right and it hurts the developer user experience like we can't compromise on it because, like you know, bank accounts and stuff can't be like exposed or like we're.

A

There's there's important secrets right, but at the same time maybe the developer experience doesn't have to suffer so bad. Maybe there's a way for us to think through how we can go in and again use something like this to provide a happier path for developers.

A

Sorry, I totally acknowledge I'm on my soapbox again, but that's uh that's kind of how I think it through um joe said this is the point of corey quinn, saying no one will care about kubernetes in the future, um not that it won't be around, but it'll be ubiquitous exactly joe yeah. You know, and I I think to some degree- that's fair. I again and I don't want to lean too much into the notion of like in order to be successful with kubernetes your developers shouldn't know it's there.

A

I definitely don't believe that that is a universal truth. I think that's just something that fits in a tweet, but I think for some folks that sentiment is totally something to strive for, if that's the right use case for you, depending on your audience, depending on what your customers need, you know keep in mind, there's always trade-offs to every freaking decision you make, and I would say that the abstraction spectrum now that we've coined that term. uh It has just littered with trade-offs all over the place. You know but yeah.

A

I think we're trending in that direction of there being less and less onus on direct interaction with cube and more interaction with some of these abstractions and, as noelle said, being the new, the new os. um Okay. So.

A

Rich said not that, of course, yeah it took longer than people thought now, it's totally fair. I mean you know, don't get me wrong, like um I would say, a lot of organizations today, unfortunately have a kubernetes strategy which might seem weird coming out of my mouth and maybe will get me in trouble but like what I'm getting at is like it's not about a kubernetes strategy. It's about a modern application platform strategy or, however, you want to coin it right.

A

It's like it's about delivering an insanely, good developer, experience strategy and if kubernetes happens, to be one of those building blocks great, but you know it might sound silly to some of us, but genuinely a lot of folks. Think kubernetes is a heroku and if they just deploy it, it will change the world without a realization of how it is just a building block and how there are so many uh steps we have to do. On top of that. So sorry, your comments are just getting me excited and I'm just talking endlessly at this point.

A

um Okay heart, you said: what do you mean by a kubernetes strategy? What I'm trying to get at and I'm not trying to make I'm not trying to make kubernetes sound bad at all, just saying, like organizationally, when an organization thinks about deriving business value. The notion that, like a kubernetes strategy, will get you there in my mind, is just it's too short-sighted, um and this is largely just what all of y'all are saying just in a bit of a different framing.

A

I think, which is like your strategy should be about having a really robust platform that enables developers to be successful. In most cases, there's obviously a bunch of cases, and if that's your strategy, perhaps kubernetes becomes an implementation detail of that strategy.

A

But you know- and I talk to a lot of customers like- and um you know a lot of times- I'm supposed to be helping them be successful, with cube, not talking them out of it, but a lot of times it's a bit of a misstep, because it's just again a poor misconception of what kubernetes is actually going to solve for you. If you're trying to save cost, I hate to break it to you, but implementing kubernetes and learning.

A

It's probably going to cost you more initially, because you're probably going to run kubernetes side by side with all your other stuff, you're, probably going to blow some things up. You know it's just like it doesn't. It doesn't play out like this thing where okay kubernetes is up in our organization. Now and now our problems are solved, which again, I know is probably quite obvious to a lot of y'all, but it speaks to this notion speaks to things like uh waypoint.

A

It speaks to how a lot of these tools are really interesting, because it's going to enable us to deliver better experiences to our customers, which, as infrastructure people as devops people oftentimes. Those folks are our developers themselves right, um cool and har. You said, but it makes you robust if you use it efficiently. Yeah I mean I. I hope it's clear that I'm not trying to say kubernetes is bad. I'm just trying to say it's. It's not the whole picture, it's a piece of the puzzle or how do people put it joe and others?

A

They say it's a platform for building platforms like I'm just saying that way.

A

Less elegantly is really all I'm trying to say um rich said yeah, I'm one of those folks who thinks dev should be working on their apps to deliver business value and learn the ins and outs and not have to learn yeah same same with me, rich, like I'm open to development teams who have you know knowledge in cube, and it might not make sense to like try to fully abstract it and make their life harder, but like um there is a large development audience where, if we can just give them a heroku-like experience, they can focus on business logic and their app just shows up.

A

That's that's pretty dang valuable and when you go and you talk and justify your platform initiatives to executive level, folks, that's what they care about too right. um Although we could try to tell them how cool the kubernetes scheduler is. I have a weird feeling. They won't be that impressed um all right and then nip and you asked uh napoon, you asked question waypoint dev: can you customize the deploy phase with a specific deploy flow like spinnaker, and I think newport I'm going to let them answer, but I think the answer is yeah.

A

I think today you do that with custom plugins, I'm pretty sure that's kind of how it would work, so you could define some plugins, probably abstract, some values that you'd be willing to take into your plugin to you know further define behavior, and then you would just kind of implement that plugin, so um cool all right all well. This was really really awesome. um Aside from maybe a little bit too much soapboxing, I learned a lot of cool stuff, I'm pretty impressed with where waypoint's going, even just as a concept.

A

I think this is a really interesting view on um how we can deliver experiences on top of things like cube on top of things like build pack and all that good stuff. So if you're still with us be sure to give a thank you to evan and cody and anyone else who's here from um who's here from hashicorp. Thank you so much for hanging out helping get us unstuck. I can guarantee you. I never would have figured out that load, balancer thing without you all being in here, so in great work, hashicorp team.

A

I hope you keep this up and keep trucking, because uh this is. This is really really cool um for us to kind of add to our our tool belt of options. So that's it for this week, thanks so much everyone. I hope you have a killer weekend and hopefully we'll be back next week with a new fun topic, maybe we'll rope joe or someone new into the into the mix to uh to deliver something so until next time all have a great weekend. Thanks again.

A