VMware Open Source, 18 Sep 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Octant - A developer-centric web interface for Kubernetes - Webinar, Sept 18, 2019

Description

In a perfect world, you would deploy applications to clusters and nothing else would be a problem. In the Kubernetes world, it isn’t that simple yet.

Because Kubernetes workloads are composed of multiple objects, even simple actions can be complex.

This complexity is why we created and open sourced Octant, a developer-centric web interface for Kubernetes that lets you inspect a Kubernetes cluster on which applications reside.

To help a developer better understand the state of the application running inside the cluster, Octant’s dashboard allows you to navigate through your namespaces and the objects they contain. It lets you visualize the relativity of objects and resources. Unlike the Kubernetes Dashboard, Octant runs locally on your workstation and uses your Kubernetes credentials to access the cluster, an approach that avoids a whole class of security concerns.

Check out Octant here: https://github.com/vmware/octant

In this webinar, Bryan Liles, Team Lead for Octant, will go over what Octant is, how you can use it, and how to get involved.

Bryan Liles is a Senior Staff Engineer at VMware. He leads the Developer Experience group, which creates solutions to help developers be more productive in Kubernetes. When not working, Bryan builds and races cars and drones.

A

Hi everyone and welcome to this webinar here for octant and open source project that gives you visibility into your cube relates clusters.

A

My name is Jonas Roslyn I'll be your host today with me, I got Bryan Liles, who is the team lead for the octant project here and he will go through the project with you all this is being recorded, and if you, if you have any questions and you can either click on the raise hand, button or you can type your question into the chat window and I will read it out aloud to Bryan during the Q&A session Bryan. Do you want to share the presentation with us? Yes,.

B

Let me get that started here all right.

B

All right so today, we're going to talk about often octant is a new open source project that we've been working on here at M, where and the purpose of the project is to provide some visual abilities to explore your kubernetes clusters.

B

So, let's get started, oh and by the way, I'm Bryan Liles I lead a group that we are thinking about developer experience. How do developers fare in the cloud native space?

B

So this is just one of the projects that we are working on, and this is the my current team I have Wayne and Sam, who are also on the call, as well so back to often what we did with what we did when we about a year ago, about a year ago, when we were thinking about projects in the cloud native space we instead of thinking of products, we actually thought of things that we would want to solve our problems.

B

We want to solve for users and a big problem that we were thinking about was as a developer or anyone using kubernetes. How do I know something is going wrong, so what octant actually started out was was a troubleshooting tool. I have some kind of workload running in a kubernetes cluster. Is it working correctly? Can I actually figure that out and can I can I show a user of my software if their software is working? Yes, I think they think it should be, and over the last year, what's happened with octant?

B

Is we've been first of all figuring out what the core was, because kubernetes was not made to do some of the things that we were doing with it, and also we spent some time trying to make the user experience a little bit more robust, because just having a view of things in a cluster without context is not super helpful. So this is what has actually evolved from answering the question. How do I know if my software is running correctly, so I would actually like to get into demo super quick and and make this super technical.

B

So let me go ahead and do that so um octant is actually a binary and and I actually I have the version of octant that we're getting ready to release today and I'll be showing that to you, and one of the reasons we created octant running outside of the of the cluster is because we want to make sure that we don't run into the class of problems where you have a piece of software that runs in your cluster, that has access to everything and then trying to pare down that access.

B

We're actually going the opposite way. We take your credentials and we interact with the cluster with your credentials, so there's never a chance that we can have privilege escalations because it's your privilege. So this is why octant is outside of a cluster and ultimately, what we would like to do is to direct convert this into a more operating system, friendly application. So right now, you'll see in a second that I'll launch octant from the command line as a CLI tool and then interact through a web browser.

B

But really the octant feels like it should be an application and it should be able to take Kay, take advantage of things that come with your operating system, such as the status bar or your system trays, depending on whether you're on Mac, Linux or Windows, and that brings me to the next point: octant works on Mac, Linux and Windows. We are trying to provide the same level of support across all three platforms and so far we've been pretty successful. So, let's get into this demo of optin and but before I get started.

B

Let me launch a web browser because when it comes to kubernetes generally, the kubernetes piece is super interesting, but the thing about kubernetes is it's there to solve problems for us. It's our vehicle for the journey. Kubernetes is not the destination. So really what we want when kubernetes is, we want to be able to run our applications and I just happen to have an application. I worked really hard to find someone else's software.

B

This is actually a piece of software called HTTP bin and, if you're doing any type of web development- and you want to test things out- it's actually one of the best piece of software out there and the reason why is because it supports all of HTTP. You can test out post, you can test out cookies, you can test out, gets, and it's just a nice simple piece of software that is great for testing. So my deployment to my kubernetes cluster is this HTTP bin app and just to show you I'm, just gonna, actually.

B

Just show you this namespace and we'll get to see it in a better way. Is this is what you get with with kubernetes. You can do a get all and you can actually look through it and if you understand exactly what was deployed and how was deployed, you can make sense of this. But in a lot of cases someone else deployed the software and someone else is responsible for taking care of it. So with octant. What we're going to show you is that it does make it easier to see what's going on inside cluster.

B

So let's pretend it's Friday afternoon at 5:00 p.m. and that one person on your team says no, let's get this change out for the weekend and they do a deploy. So I actually have a deployment here, we'll just call it release and I'm not going to show you the yeah Mille. It's not really important for this piece, but we did a deploy on a Friday at 5:00 p.m. and what happens on a deployment at Friday at 5:00 p.m. well.

B

It breaks and now we're in the place of well. What do we do when our previously working or never working, think our application and and kubernetes is not working? Well, if you are a kubernetes expert or even someone who's, just a little bit familiar with kubernetes? What's the first thing you're going to do well?

B

First, you need to know what workload it is. Then you need to know how its configured to talk to the external world, and the problem is- is that we assume that you have some level of experience that isn't really required for application developer. So this is really why we invented Achtung. So without further ado, a labuda baktun, like I, said before it's a binary and it runs as a web browser application right now, so I'm going to launch auxin and actually I will make you just a little bit wider.

B

So here is often- and this is currently as the VAR 0.7 release- that we're going to do soon is the home page, and what we're showing here are a few things. It's actually just a dashboard on the left is all the things you can see and on the right is a expanded view of whatever you've selected and the reason I'm being big is because, depending on what you've selected is what you'll see so right now we're looking at an overview of the cluster and our of a namespace in the cluster is 2019.

B

Oh eight, nine, nine one eight and we can see all the objects that we think are important for here. So, whether they be workloads or or secrets or or custom resources, we can show them all same-same pain. So right.

A

Yeah, just a quick thing here: can you zoom in a little bit as I, can.

B

Thank you so much yes. Actually, this is what I'll do I'll make it fullscreen and there we go so yeah. Just let me know if it's the right size, I think.

A

You can will one one zoom step in perfect. Thank.

B

You alright, so now what the problem we're trying to solve is website doesn't work. So what do you do? So? This is an interesting problem in itself, because there's not an if you're, not an expert or know how your application is configured. What do you do first? Well with optin? This is what we've done so right now. What you can do in ogden is. We know that this is a deployment of some sort, but do we know exactly what's deploy minutes?

B

It is well I can guess, because it says HTTP Ben is one of the name of my deployments. We can actually do that, but there's actually something better, that we can do inside a baktun.

B

We can actually filter based on labels and the reason why you'd want to do this is because a particular workload in kubernetes or an application might have a deployment it might have a service and an ingress might have more than one I'm config maps or volumes or secrets there, depending on how many actually applications are running, there could be different service accounts and there can be custom resources. So what we did is we actually put labels into octant to allow us to zoom into what we want to see and the reason I bring.

B

This up is because the kubernetes in the kubernetes project we worked really hard to define a set of labels that will allow you to figure out what an application is so I haven't I, don't actually have them labeled with that standard, but I do have this project H been label that we can actually use to to what we're seeing. So what I've done here is I filtered.

B

My view now to just only show things that are in this project: :, HTTP bin and what you'll see here is it got rid of a lot of things, but now we see that in our deployment there in our workload or application, there's a deployment and ingress a service and and some kind of custom resource, and so now we have even a smaller surface area to actually figure out. What is the problem? So, let's go back to our error.

B

It says they should be been page can't be found. No web page was found for the web address. Well, actually, what does that mean here? So we know that ingress is responsible for serving up websites or serving a basically hosts inside of kubernetes.

B

So let's go there first, and what I'm doing here is we'll just go to this HTP bin ingress and we'll see what it actually says, because what we're looking for are clues to what the problem could be and ultimately, what we want to do with auxin is not is instead of having to look for the clues, we want to actually tell you what the problems are and then give you the easy fixes.

B

So, if we're looking through this, we can see that there we in octave, we have a summary view that we've gone through and we this is still bound ingress to its important parts. So for every single object inside of octant you'll get configuration sometimes we'll get status. You'll get the metadata you'll notice that I applied this with cube control, and because this is an ingress, we can see our rules and we can also see the events for an object. So if we go through and look at the events, let's see what we see here.

B

The first message says error: while valuating the ingress spec could not find service, and it says the service name is HTTP bien. Well, that's obviously a spelling error and you can actually see that right now with cube control describe, but the problem is that you would have to know how to run cube control describe so we put another feature in octant, which was the original feature. How do I know what's wrong? So what we have is this thing called the resource viewer and what it does is.

B

It looks at an object and then then it looks at all the objects that our object is connected to, whether directly or indirectly- and it goes through that entire tree or that entire graph and it actually builds up the state and what we can see right now. Is that a should be then it this ingress is in red. It's an error state and what öktem is telling us is that a back-end refers to service that doesn't exist, so that is the first clue of what could be.

B

What is happening here now, if I go back to the summary, what we would like to do is have a way to edit our rules here, but oxygen is still young projects, so we don't have editors for everything, so, for a couple of these I will go to the command line and actually edit them from the command line.

B

So what I'm going to do is actually I'll use a cube, control edit and what I'll do is go down to my typo and what I'm doing right now is I'm editing a basically a manifest. That's on the cluster and I'm going to go fix, my type of so we edit this and you'll notice it. It actually happened, pretty quick in real time what octant picked up the new back end now and what we're seeing is events coming in real time and- and this is the power of auxin- it's a real time tool.

B

You don't have to set polling periods or anything like that. We've set it up so that when things are changing that we're telling the front end whatever the front end might be what's going on. So if we go to resource for now, we noticed that hey. My ingress is now connected to a service, but my service has my service. Has an issue now and since we're still triaging this problem, we'll say all right.

B

What was this error service has no endpoint address, so in Coburn your workload ultimately creates a pod, and what happens is whenever you want to send traffic, whether it's in the cluster or out of the cluster to a pod, you would use a service and what, in the way that the service works is that it matches labels from the pod to a selector. It is defined in a service. So actually let me show you how that actually works.

B

So inside of the amyl, we have this thing called a selector and in this case it's run HTTP Ben. So what it's going to do is look for pods that have the same, have the labels that matches selector and then it'll been the it'll, build up these endpoints and that's how it routes traffic. So what we've noticed here is that says: the service has no end point and the reason the service has no end point is because well actually I, don't know why. Let me go look at the pods and and we'll load this.

B

So what you're noticing is that it says that we have no pods what auxin is doing in real time, because the kubernetes api doesn't have support for this type of interaction. Is we're building up these query, things called Informer's and what we're doing is for every single type that we want. We connect this and for more up and what it does. It watches a particular resource and the in the in the cluster. So in this case, what what it'll be is looking at pods.

B

So let's actually have a look at our pods here and wonder why we don't have any pods loaded. So oh I know why this is the reason. Why is because I didn't set up my deployment to propagate my labels, my bag? So when we go down to here, we can actually see that we have a pod. That's called HV Ben, but we noticed the label on.

B

It is a PHP Ben and if we remember earlier back on our service, it was run so now there's an issue, and let me show you how we would fix this with octant, so we go back to my service and we'll go to this service again and let me go down here. So if we look at the service, this service is set up to talk to this selector, and this is the wrong selector. There's actually nothing connected, there's no pods that will answer to the selector.

B

So what we want to do is we want to be able to do this inside of optin. I want to be able to edit and not have to use cube, control, describe or get or edit or any command, because we want this tool to make cube control fit in the graphical world. We don't want to replace it, but we want to make it better. So now, what we're doing is we can actually just go.

B

Look after all, the selectors that we have- and the editor here is very simple and we're going to actually just change it. We can necessarily choose multiple ones, but we only want one so now we're changing our selectors for our service to a chap, :, HTTP Ben, and what you notice now, there's an endpoint that now shows up and it matches our one pod and the rete, and we can verify this is working now.

B

Well, our graph that looks way different and now what we can see is that we can see that we have an ingress and our ingress is connected to some kind of resource. It will just connected to another custom resource. We have. An ingress is connected to a service which is connected to one or more pods, which is we're connected to a replica set which is connected to a deployment, and the pods are also connected to a service account, and we use this. We use this information. Determinate things are working correctly.

B

So if I go back to my original page and I hit reload, it shows up, because now we have this graph and we in the way that we know that it works, is because we can verify it in in real time. And this, let me show you another feature of octant here and and show you this whole real-time thing.

B

So what I'll do is I'll edit now I wanted that I'll edit, the I'm gonna edit the service just by hand real quick just to show you that in real time that this tool is actually doing that, so we'll go back to run again and what we'll see in real time, no changes that it actually does all the things here. So we'll edit we'll change it back now do what we know should work and.

B

Oops there we go and there we go so now our site is working. So what I'll do for the next couple of minutes is I'll walk through other features that are inside of auxin, it's more than just just showing items in a dashboard, and we have this resource viewer. So before I was saying that we have yam. Oh, this is just a yam, oh that or is this just the object in this case the service called HTTP bin that we rendered as yam?

B

Oh, so you can actually see the configuration because there's items in the summary that trans don't translate well to our our components that we have here. So we want to make sure that you have everything, and specifically those would be things like some of the status items. So service is a bad light, is a bad one, but we can actually show you show you everything lose how the cluster sees it. So you'll notice that we have summary resource, viewer and gamal, but let's actually go check out some more some different objects.

B

So let's look at a pod, so this is the pod that is serving the container. That has our website there's a couple things interesting on this page, so same thing, so we have summary resource, viewer and Yambol, but another neat feature that we can add here per type. We can add other tabs so for force for pod. We also might want to see logs because logs because containers have logs and you could use the cube control logs command to get this out.

B

So what are the logs look like well they're, just the logs that the process started. So this is a very simple log viewer right now we're reading it right off the API and we can show it with or without a timestamp future featured ideas, for this page is we're going to add support for different types of logs, so I know I can borrow LogMeIn Jason, and we should be able to support that here and another thing that we're going to add is being able to see logs across multiple pods at a time.

B

Would it be interesting if you were running more than one pod, so you don't have to actually go search, other things that we have. If we go back to pods our port forts and if you can remember the way to start a port forward, it's cute control port dash forward and then is it the then when you type next, so maybe it's a it's a service, the deployment or the pod name and, and then this is local port first or remote port first! Well, you know what I don't want to remember any of that.

B

So now, if you have a service or anything Armina, if you have a a pod with a container in it, and the pod actually defines this container ports and it's a TCP guess what we can give you that right in a browser. So what we're doing here is I'm starting the port forward and I'm starting it on your localhost. So it doesn't matter where the cluster is running. I can actually just show you locally.

B

What's going what's happening on that port, so in this case, I've actually started a port for on localhost six, one, five, three four! So when I click on it we get the same view, but this is actually using the port forward. So if I was, if this was a remote cluster, I could actually see it like it's local.

B

One thing that we're going to do for this is that, right now we only support services. You'll notice that cube control supports, like I, was saying before it supports deployments and services as well, and the reason that we're not is because of the way the cube control. Does it it's and determine what happens.

B

Is it says, give me a service, are giving me a pod for a service or give me a pod for deployment, and you don't under and there's no way inside of the kubernetes code, the track that what actually is getting port forward, because you're, not port, forwarding to a deployment or service, so we're going to work on getting this in the tool, so we can actually track it and we want to be able to track it actually elsewhere in the application, and I will go there right now.

B

So what we can do is, since we have the port forwards, we can track all the port force you have going at one time.

B

You can actually have multiple port for it's going and then you should be able to control them from one location, and this will be super helpful whenever we do it with services and deployments so moving on here, there's other interesting parts: we've broken up, octant views into multiple main views, so there's an overview, a cluster overview and a configuration so inside of kubernetes there's namespace scoped objects, the objects you think about are like services and pods and and and service accounts and config maps, and then we also have cluster overview.

B

Those are these are the non. These are the cluster scoped objects. So and in our case right here, you have some are back where, whether it be cluster roles or cluster role bindings, but you also have things like nodes and another thing you would have: are your cluster scope, your custom scope resources? So what we can do is we can show you this, and this is actually another interesting tool or a piece of octant that we don't see out there.

B

A lot is first-class support of custom resources with chip control, um I think ever since kubernetes 1.12 might be a 1.11 or 1.12. You were able to have custom printers and because you could have custom printers, we could define our custom resources and we could create custom printers. So what you'll notice here is this is a list of of cluster issuers. Octant knows literally nothing about cluster issuers. All it knows is that it's a custom resource and it doesn't have a custom.

B

It doesn't actually have a custom columns, so all octa can do with it is say: well, I know it has a name. I know it has. Labels and I know it was created at a time. So whatever Aachen does this, what we can do is we can kind of guess it. What we think is actually in the object. So we don't know anything about config, but we can definitely show metadata and that's kind of neat.

B

But a very neat piece of octant is that we can, if we can show more information if the CRD actually has more information, so you'll notice that this custom resource for certificates has more columns. This none of this is baked in octant, it's actually based baked in the CR D. So what we can do is take the custom column information and we can apply some logic to it and now what we can do is generate.

B

We can generate views that have a little bit more information in there, so we can figure out kind of what's configuration or it's kind of what statuses. So what this does is at just out-of-the-box.

B

It allows optin to have this neat information about about custom resources and see our DS, but we can do one thing better and let me actually show you this I didn't turn it on when I first boot up the app, because I want to talk about other things first, but what I'm going to do is I'm going to install an octave, plug-in and I'm not going to tell you what it is before I do it, but there we actually have support for extending optin like I saying before. Ahkna can only know about things.

B

It knows about and see our DS, our customs there's no way that Achtung can know about C our DS. So what we can do now is we actually have a way that we can extend octant without extending the binary itself and what I did real quick was I installed a plug-in, and this is just a sample plug-in, and but it does is a few things first for for pods what we can do is we can extend the information that we're showing so before when we had a pod.

B

We had summary resource viewer Gamlen logs. This plug-in is an example plug-in, so it's actually just showing the power of the plug-in interface. Now, what we can do is we can. Actually we can change configuration so from the plug-in we can make. Real-Time updates and I made this a number I made this an increasing date, so you can actually see a change. We can actually change status as well. So we can say that there is extra status, that I know from my custom object or from an internal object.

B

I can print that as well and have it show up in opt-ins interface and another thing that we can do is we can actually just add extra output, so we can add extra output from my plug-in. That makes it look like it's built into octant, but this is coming from the sample plugin that we just installed and likewise I was showing up here earlier.

B

We can do things like with exterior tabs, we could just put content and this content can use the same widget set that everything else is using, but it's running in its own external process and another thing that we can do is, we can add different. We can add a whole new menu. So what we did here is we just added a whole new menu.

B

So if you had a service where you were doing something, like is custom CD some continuous delivery, we could integrate this tool right here in the auxin, so we can show we can show your Tools output, but we can also show it in the context of what's happening to the cluster and I think that's missing for a lot of people. So in this case we just happen to have an extra tab that we put in the plug-in and the tab can change values.

B

So just a very simple example, so Austin is built to help you see what's going on in a particular cluster, but it's not limited to what it knows we can in belt. We can actually make that knowledge much, for we can make it much nicer with this plug-in interface.

B

So, let's see what else we can deal with optin, and this is actually stuff. That's coming in our our next release, right, 0.8 release. So let me show you some more a few more things about octant and then we'll turn it over to roadmap and Q&A. So we want to be able to not have to use cube control. Edit I mean this is only if you're in a development cluster you don't have to use cube control edit. So what we've done is we're experimenting now with editors.

B

So for a deployment, for example, I want to be able to scale this deployment up, so what I'm going to do, and actually I'm gonna change it to a different deployment, so something I actually know the the docker image names for and first thing I'm going to do is I'm going to scale this engine next deployment up to five replicas and what's happening now, is in real time what happened?

B

My replicas and the neat thing about these menus is that even if you were doing you can actually add these types of menus from plugins, so we're adding basic editors, but you could actually change this editor to edit more complex things and there's also other things that we can edit as well. So in a pod in a pod template for any deployment, there is containers. So one thing we can do is we can say that this is our app. We wrote engine X and we want to actually change the version of engine X.

B

So now we'll just change it to okay, 1.15. So we'll say that we've moved engine X to 1.15, so we've in from the from the actual app we actually were able to do this and any part about this is that it's down to the bottom. So we can see in real time what the cluster is doing. So it's scaling down a replica and first it created a replica and then it started moving one object from one to another, and what we can see here is if I go to the resource viewer.

B

Everything is up, but this replica actually is brand-new and to help with that view. Another thing that we're going to do- and this will actually be part of this- will be in our next release as well- is that we want to be able to. We want to change how people are looking at kubernetes clusters. We are so used to looking at dashboards. We default there, but I think that we could actually, if we could rally around a view like this put.

B

The good information towards aside, you know have simple views, have simple editors, a lot of the work that we can be that we would do with trying to work with our kubernetes cluster can happen from a screen like this. So I could have a deployment editor right here on this screen and in real time we can see how it is changing my cluster and if it breaks, then we can have the alerts come right here on the screen same time, so this will is actually. This is the yeah.

B

This is where we'll end for for this bit. There's other views in here, so every tick, every every type that we have in here has its own view. Besides custom resources and what we're working on now is now that we have a foundation is making these views better and making them more useful to to our users, and we also are listening to our users. If someone came and a couple weeks ago and said, I want to see nodes so node show up, and now you can actually get a view similar to what you would get.

B

If you did a cube control described node, it's not a hundred percent because we haven't put in pod yet, but that will come pretty soon, where you'll actually be able to see the pods that are running on a particular node. So I will stop demoing. This now and I will go back to slides than me.

B

So, oh. The next item is our roadmap. So before I talk about the roadmap, let me talk about what was going with actin in our next release. We are preparing a 0.7 we're octant 0.6 now, which is our latest release, we're preparing a 0.7, that's coming very soon. We're almost there if we don't define anymore bugs and the difference between octant, six and option seven is 0.6 was our first public release. It's where we showed it to the world.

B

um 0.7 is a lot of feedback from a lot of people where they are finding bugs that we wouldn't find things like how we were Hamlin, CR DS and it's nice having the diversity of people like a huge diversity of people. Looking at this, because there's lots of corner cases that we didn't see so and 0.7 what we've done is we've actually made multi cluster useful, so something I didn't show. You was we're on the cluster that I have I, just call it GK cluster one.

B

It's just a cluster that I have running on on gke. You can actually swap and I might've config isn't set up correctly, but you can actually swap if you have multiple. If you have multiple cute config locations, and then you have multiple contexts, I couldn't can switch swap in between this. So that's a that's a pretty neat feature if you're actually have to work on multiple clusters or if you work on a cluster where you have a read-only view and then you're moving to a readwrite view for safety. So that's that's a good feature.

B

The biggest feature of between 0.6 and 0.7 is how we do the backend.

B

If you notice this all, looks pretty seamless and but there's a lot of stuff happening in the background to make these views show up correctly and before we were using HTTP long pole to get the information, but we're finding that, if you do that, you have to keep a lot of data in the front end and I've showed you a web GUI, but there's no reason why there isn't other types of front ends for this. Maybe integrate it with your favorite editor or integrate it somewhere else.

B

So we've swapped the back end to a WebSocket back in it allows us do you have a synchronous communication, but it definitely allows us to move all the complexity and their front end to the back end, where it's actually easier to maintain, which makes it easier to write friends. So that was a huge change in this in this segment and then we also added editors moving forward.

B

We actually do have a a roadmap for octant and I want to highlight some of the important pieces that I think so the biggest one is viewing and editing your kubernetes objects. I want to have the same fidelity or even better fidelity than editing objects. Then you get with cube control. Editing, as you saw before, I did an edit and I ran up the command line. It launched my editor, which was them, but you had to know what to change.

B

I want to be able to show you what you can change and then give you inputs for that change. It possible. Another thing that we want to do. Is we realize that the my editor is important, but we want to be able to make that a better view, so in the editor I want to whenever you're looking at the editor I want you to be able to see things like we have pod cider. Well, what does that mean in the kubernetes api? There's Doc's, there's a lot of documentation for what this pod cider is.

B

We should be able to show you that in real life, in real in real time, in line other things that we would like to do is extend the plug-in. Api I showed you a little sample of it, but we wouldn't actually make it robust. So people can write cool things.

B

On top of this, google has a project called Tecton, which is a kubernetes aware, continuous delivery tool and they wrote a plug-in for octant and I want to be able to enable a lot of people to write plug-ins for this, because we are more of a journey tool.

B

We want to be a part of whatever you're doing, but we can't know everything about it and and in some certain situations like with your editor use, vs code, you don't need a go extension if you only write Python but because villa's code has the great editing experience and the great extension experience. We can actually provide that feature and it works, and we can do the same thing with octant and another item here is speeding up.

B

Octant I mentioned before a little bit about how we're using informers that are in the kubernetes client to actually talk to kubernetes. It's not real-time. It looks kind of real-time ish, but we do a lot of tricks to make.

B

It feel real time and there's things that we can do to I think we can do more and often to make it even feel even more real time and reduce memory usage at the same time, and the last piece here- and this is the most important one to me- is I- want to make sure that this is an application I wanted to make. It feel like it's part of your desktop when it's in a web browser it's interesting, but we want to make it feel like it's a real application.

B

So with that, please take a look at the rest of the roadmap and what I'm going to do now is if there are any questions, I'm here to answer them for you.

A

Awesome, thank you so much Brian. Do you want to put up the the slide where people can get involved with this yeah.

B

Let me do that. Let me share my screen again here.

A

B

Right so share this one go back here.

B

Yes, that reminds me before we get into questions. I do want to say a couple more things. This is where our project is where VMware / octant on github, we have a Twitter and we have a kubernetes slack, and someone from my team is always there.

B

So we are there answering questions and helping people get started and we do have a mailing list, but the most important piece is: if you want to be involved in this community, we are trying to build it and we have a meeting on every second Tuesday right now and I think the next meeting that makes it the 8th of October.

B

We are definitely accepting pull requests. This is not just a project that we open the source for no, we are accepting pull requests from people and people are sending pull requests. So please come look at what we're doing and now. I really will end.

A

All right awesome, we do have a few questions here. The first one is from Eric DeWitt. He tried octant from a customer laptop today, cube CTL is connecting fine to the cakes cluster. He started. Often the web portal opens but doesn't show anything, does often use a special port to connect suitcase or as there any laws he can check.

B

So I would most likely guess you're using octant 0.6. The 0.7 was the I would almost be short of whatever you are running into and there's a few things that you can run into, especially if they're running CRTs that you could have ran into our next release should solve that issue because, right when we released it last month, the first couple ones where I'm getting CRT errors and it's not connecting I'm, not having I'm having this issue, and what we found is that people were creating CR DS.

B

That kubernetes would tolerate, but they weren't actually correct, and what octant was is because we were so strict and how we were parsing them. We would just bail out and then not show you anything I, don't know if that was the right answer, but that's that's where we were the the real answer to this is, if you run octant with B, it turns into verbose mode and I'll. Show you a lot more stuff.

B

Please either send us on the mailing list or come into our chat or create an issue and we'll help you get through the problem.

A

Excellent, thank you next from Tom or dumansky any chance that octant will support multi cluster connections in the future. Yes,.

B

There is a huge chance that ox it will support multi, clutches multi cluster connections in the future. So the reason it doesn't now is because that was not easy. It's an easier problem to get started by connecting to one to make it support multiple. We were actually going to have to change a bit of our core, because this the way that we we work with a what we work with it. But there is nothing there's, no technical reason why we can't do this.

B

We're gonna try to figure out when we're going to get it prioritized, but it's definitely on um on my mind and Wayne's mind right now,.

B

A

C

B

Answer is, you could just run two instances of often now you have to change the port, but that will get you through right now, but we want to have a better experience than that is very.

A

Alright, next from Eric cardi, well pb's reviewable in the future.

B

Well, pb's be peaceful in the future, yes create an issue, the we what we did was and when I started, this I did a survey. What was going on when I just took a survey of what, what a few you guys were doing for a dashboard and we picked the most popular ones having TVs viewable, yes, and also someone else earlier today reached out about network policies, look try to make them. Tvs will be easier to actually show, but we'll try to get those those objects in.

B

But let us know whatever else you want to see now and then I'll just read the next question, because I have an answer for it and some re talking Eric asked again. He says this octant run a night. Kubernetes cluster will would be able to view the corresponding resources and octant. So when I first started, this I mentioned that we prefer to have optin run locally with dear local credentials.

B

That does not preclude you running octant in a kubernetes cluster there. If you we actually have a docker file, you could actually build a docker image. We're not we're not supporting it right. This second and you could deploy, you could create a deployment and you could deploy octant to your cluster I've done it for an internal team and it does work there are. There are a couple of gotchas that we're working on now or we're trying to work out now, but yeah. You actually could run auxin inside of a cluster and share with multiple.

B

Glad Jonas awesome.

A

Yeah next one here- this is just a little more regarding the octant logo was an octagon considered instead of a heptagon.

B

You know I write software I, don't know the real answer for that. We we yeah I, don't know the real answer, for that sounds good yeah.

A

I've moved to the next one can octave. This is from beeping at jet. Hwanhee can octant run on a headless Linux instance and I use a browser or another machine to access the GUI. So.

B

um I gotta see how we connect, we create the listener, um but it shouldn't be impossible. If we can't do it, it's a one line change to make it work. If we could but I, don't know out of the box. I don't know if we I think he looks at the port, but we don't want you set the host and we pin it to localhost, so it's possible I'm. Actually um it would be great if you created an issue for this.

B

That would be that's way, so we don't forget it and we can get a prioritize, but I don't see a problem with that, especially if you want to run it on a headless Linux instance.

A

Awesome. Thank you. Next one is from Vladimir Vivian storage. Visibility by a CSI would be great. Are there any plans to incorporate that.

B

We don't have any plans to, but and I will say the first thing: I am NOT an expert at CSI and I, don't even know what we would show. So what we would need to do is it's the first determined we want this in core. We probably could put this in court because it's CSI and we would need to collaborate with somebody to help us understand what that B would look like and so I'm encouraging you lad Amir to reach out. We can definitely do that.

B

A

Thank you and then the last question here is from during Moffatt it's a non-technical question regarding hey Jeremy I can take that offline with.

C

A

So just reach out to me on the kubernetes lac and we can get that sorted.

A

Let's see, I think that's dead, yeah, no, more questions here in the chat or any raised hand. So I think that means that we are done. Do you have any parting words Brian I.

B

Want to say that I am super happy about all the great feedback we receive from this project. Own gel, beta and I talked about this I'm Joe bata, the creator of kubernetes. We talked about this about a year ago, like literally a year ago and and I thought it was gonna, be a good idea. He went off and did other things like sold. Hefty o2, VMware and I worked on this and it's great to see that.

B

But but people are seeing it from the way that we're seeing it and we're just getting started on some of these amazing things that Achtung can do and I really think that if we get it even halfway, right will change how organizations and and groups of people are adopting kubernetes because we're making it accessible and so I'm I'm. We are so at the beginning of this right now. It's crazy.

A

Awesome, thank you so much. We actually do have one more question from Vivian's. I'm gonna allow him to talk here.

A

Libyan. You can unmute yourself.

C

Alright, can you guys hear me? Yes, yes, so quick question, um you know being in the industry for so long I've seen, tools like this that has grown up to be very you know, adopted all the places now the question I have is: do you foresee octant doing things like allow you to deploy not just the reporting tool but an active tool that allows you to deploy resources onto the cluster as well?

C

B

C

B

That's why we created the plugins. We never want to do that. I see. um We want to be a view. Tool of understanding, kubernetes objects how they get there. You might have an editor we might put in a feature actually I prototype one. The other day where you could. You could actually go to often say I want to create I want to import a manifest. We will load the manifest for you and then disseminate through it, but the objects out.

B

So you can actually see what you're putting in your cluster and then before the 1.7 are the 1.16 server-side apply so for 1.16 and 1.17. We don't even know if server-side apply works. Well, we can do something where we can do a simple thing: where we'll apply, we can apply your objects and we'll just do CR DS first, but that's about as much as I can ever see.

B

We would actually rather integrate with figuring out what helm support for auxin looks like or working with, like tilt or garden or scaffold or Tecton, or any other other married tools that we could think of. Okay, I'm excellent I'm, not trying to put everyone out of business just just.

C

Some right, some of you, alright thanks man, all.

B

Right, no problem all.

A

Right, I think that's that's about it and thank you so much Brian for the presentation today really really a appreciated and thank you so much everyone who joined in for for the webinar. We will post this up this up on YouTube later today. So if you missed anything, you can watch the recording. You can share it with your colleagues, friends and family, and with that, thank you all and have a wonderful rest of your week. Thanks for coming have a good one fight.