VMware Octant Community Meetings, 5 Aug 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Octant Community Meeting - August 05, 2020

Description

August 05, 2020
Agenda
- Demo of Jenkins-X plugin
- Prepping for 0.15
- Fixed context selector
- Fixed multi kubeconfig bug
- Added Webhooks menu (was API Server)
- Added component stepper for plugin authors.
- What's next
- Many annoying bugs, we want to fix them. Use the next week to give dedicated attention to some outstanding bugs.

A

All right welcome everybody to the august 5th octane community meeting. uh Today we have a. um We have a few things. We have a demo uh from the authors of the jenkins x, plugin that we'll we're actually going to start with uh then we're going to get into what we're prepping for the 0.15 release, which will be coming out uh later this week, we're going to get that released and then we're going to talk about what we got coming up after that.

A

So I guess I will just uh get started right away with the uh the jenkins x demo, uh so james and um james, uh but yeah so uh james, uh stretching um james tracking is going to do a quick demo of the jenkins x plugin, which we had actually talked about uh at a uh when previous, when you all published the repo we mentioned it at the previous meeting, so uh yeah go ahead and take it away. Awesome.

B

um I'm gonna stop my video just because I don't have the best internet and I'll try to share my screen there. We go two seconds.

B

Okay, so hopefully in classic presentation song, can you see my screen? Yep yeah awesome. So this is the octane jx plug-in. um Everyone here knows opt-in, so I'm looking at this grab optim I'll just talk about the upton jx plug-in, so the jx plug-in is basically a plug-in that tries to visualize as much of what's going on inside jks as we can.

B

It's still fairly early. We've only spent like a couple of weeks on it um already it's pretty awesome uh and we're hoping to extend it to be the full ui for installing jenkins x, upgrading jenga's x, managing jk6 as well as just you know, pipelines and using it for those who don't know jenkins x, it's basically automated ci cd for kubernetes, it's based on tecton.

B

um So let's switch to octan. So here's octant it's running in the kubernetes cluster, that's a pretty empty cluster. I've just installed jackie's x version three, which is just about to be released as alpha, and you might notice these couple of icons down here. The one right at the bottom is the operations view for people administering jenkins text, which I won't show you right now.

B

It's still a bit proven ready, but I'll show you the main developer views so there's a bunch of views that are basically visualizing jenga's x, specific resources, mostly and so they'll start the environments view. 26 has a bunch of environments, development, staging production. These are places where, when you create apps, they get deployed the development environment. These are all currently shared inside the same cluster. So there's only one git repository right now, which deploys to dev staging production.

B

If you have multiple clusters, you'd have separate repositories. So this is the source code that defines all the apps that will be deployed via tools like helm and whatnot.

B

Another view shows the repositories, so these are the git repositories which have been imported into chinese x, which we've enabled ci cds, in other words, we've registered web hooks. So when you change these git repositories, ci cd happens via tecton. So here's a quick start, which is just a very simple, go microservice.

B

You can see how totally awesome it is. It prints some stuff and I'll use that in a second, then we have the pipelines view.

B

So the pipeline's view is the main breadth and butt of this plugin, where you can watch all the different pipelines running in your cluster just to make it a bit more interesting I'll edit the source code in github and I'll show you how pipeline triggers um I'll say very awesome, I'm going to be really naughty and I'm going to no, let me let me be real, I'm going to create a pull request, I'm going to create a pull request to change this awesome message.

B

That's going to appear that says very awesome, so I'm going to say, fix more awesome, I'm going to create a pull request, so I'm going to propose a change and then create a pull request. So we've created a pull request on this source repository. This is all kind of janky dexter for a second. What that's going to do is trigger a pipeline, we'll see the pipeline trigger in a second there's the pipeline started. If I go back to the pipeline view, we'll see a pipeline started here. I can filter on the repository here.

B

So let me just show just this skip repository so we'll see the pipeline has started, but we can just use this view to look at all the pipelines across all organizations, repositories, branches, bills and status, whether succeeded failed or running or whatever.

B

Now these views are okay for just seeing what's going on. If you wanted some something a little bit more interesting, you can click on that link and then it shows you all the steps that are occurring inside that pipeline. So, for example, this one I could click on this link and it takes you to the octant log view for the step that's running in tekton.

B

So I can look at all the container logs for all the steps that are going to take them and each of the pipelines are attacked on pod, and I can click on this link here to look at the pod. So if I click on this view, this is the traditional classic view of a pod in in octant, just wait for it to load. So this is looking at the pod. That's running the tecton pipeline, which is pretty big.

B

There's lots of container there's a few unique containers, lots of containers each one of these containers is a step and you can be quite unwieldy.

B

Looking at this because tectonics is so kind of complicated and the events at the bottom are really handy for when things go bad, to make things a little bit easier to understand when you're trying to diagnose a step, sometimes you're working on the tectum pipeline in jackie's x, and it does the wrong thing or it failed for some reason: there's this little uh envelope icon, it's not the best icon in the world.

B

If you click on that one, this shows you a step view which is kind of like a simplified view of what was the container doing for that step. So these are all the volume mounts that are happening inside the pod. These are the environment variables here, there's quite a few. This is the name of the step inside tecton, which is effective thing. This is the container image that ran the step.

B

So you can see ah this was the canika image from the chemical project and if I click on that link, because if it's on gcr or it's on the public container registry ecr docker hub, we can link to the actual container, which is really handy and then there's the actual command that was running. So this gives you within a couple of clicks from this page. We can navigate into the logs for a step the detail of a step uh and obviously see this step and there's a bunch more links as well.

B

If you're a jenkins x person, um we can click on the pod, as we've shown this so see, all the steps see the logs of the whole pod there's also a quick link. Let me shut some of these links because it's looking confusing, I can do a quick link to the source repository that triggered this repository, which triggered this pipeline, or I can look at the pull request directly, which is really handy so basically from this apparently pretty simple ui.

B

uh This is all implemented in a couple of panels, with some markdown and from this relatively simple ui. We can navigate to almost everything that's going on in jenkins x and in tecton, which is really really awesome. Now this is just to start, I'm sure we can make this much much nicer, as we mentioned before. This could use some of the nicer views from the clarity css.

B

We could maybe do timings a little better, there's a little bit of tool tip there, but it's not the most awesome visualization. We can maybe use in my iso widget for timing and durations um one thing: I've always wanted. It doesn't quite work yet is when the pipeline step fails, it would be awesome to open the terminal in the pod and it doesn't quite work.

B

There's an awesome terminal tab in um in octane here, which is awesome and it works for almost every pods I've tried, but for some reason it doesn't work for any of the tecton pods I try, but for every other part I try it works beautifully.

B

So being able to open a terminal in a step in a pipeline would be super. Super awesome um but yeah never mind about that, and so that's mostly kind of it. Oh, I should just show you that um that pipeline I just triggered here. If I go back to the pipeline, uh this was a pipeline just triggered and if I click on that preview link, there's a link to the application. That's running so there's the very awesome. So this is the microservice.

B

That's just been deployed in a preview environment, so I can try it out before I merge the master and if we go back to the pull request we'll see before we've merged this change to master. We've got a comment on the pull request on github and again it's a link to the preview. So we can access the preview from either github or octant. So we tried to kind of link both uis together, so you can either look at things from github or from optins um over time.

B

It would be kind of nice to do more links. So we could maybe link this pipeline view to octane, possibly using say a localhost url or something so you could. If you're running locally, you could browse from github to octant directly rather than going through a running app, but that's kind of mostly it. So we can look at repositories previews. We can look at running apps in the environment, so here's all the current apps that are running um we're, hoping to grab a bit more of the helm metadata.

B

So we can show icons and what happens your home pages of apps, here's, the ingresses for the apps that are running in this namespace and we've got a little helm, uh plug-in um in jenkins x3 we're actually kind of moving away from doing direct helm deploys and we tend to use cube ctl. So a lot of the helm metadata doesn't exist.

B

So there's only one helm release in this example, but the main ui is really about um pipelines and visualizing the pipelines and go from here to navigate into an individual pipeline or filtering the pipelines by you know, environment or whatever or or by branch like you might want to see just the releases or whatever. But that's pretty much it um any questions at all on anything.

A

uh Not so much a question, but just a just a comment. I think so. If I understand this right when you're looking at the uh the pipeline view and you're and you're looking at a uh a build the um if you, if you had one that had failed the the I guess you called it the car, the task, the the envelope icon.

C

A

You could you could take that and you could go to your local machine, pull that image into your local docker and then essentially copy and paste that command.

B

A

B

Yeah there's actually uh a couple of issues on the octane jx, so if I click on so here's one, that's failed example. So if I click there, I can see the log, that's failed, so that's the failure and oops.

B

Let me go back here and if I click on here, I get the detail of the step, so here's the environment, variables and here's the map points or whatever what would be really nice and there's a couple of issues already for this is a link on this page that says, if you want to try this locally either natively on your laptop, you know create all these environment variables using export. This equals the export basically excellent and then run that command.

B

You know just so you could run it locally or do a docker version of this spin up a docker container locally with this image with this file system and this environment- or you can maybe even repeat this by doing a you know, cube ctl, run and again same kind of thing passing over stuff. So literally it's just a couple of pages for some text. We could copy and paste it as we're developing, but just having that, because I doing a search replace on this to add the export statements of the quotes around.

B

It's just tiring, so yeah I'd like a little bit more to to make it easy to reproduce a tecton failure and and diagnose it locally and another thing, I'd love which would be totally awesome. um It's been discussed inside tecton of having a way of pausing failed pipelines.

B

So if you can get the pod to fail, you can then just cube ctl exec into the container in the pod. That's failed and it's all right there right all the vulnerables permit. Verbals. Are there the file systems there everything's right there, so you can just run on it and if you could figure out how to open the terminal on that part, which I'm sure is a fairly straightforward thing. I haven't looked too deeply because terminals work perfectly on every other pod. I try it's just. It could be a checking sex thing.

B

It could be a technical thing, I'm not sure it could.

A

Be a bashful yeah, that's exactly it! So we we just default to uh sh and if it's not there, it's not there and and and it won't work. So it's probably something we can do with either uh allowing uh the terminal to be informed of we used to have this and we took it out because it was a little confusing, but some intuitive way to say like try, this list of things or um you know let the maybe maybe let the user define what what thing to try from the case of like a plug-in.

B

Yeah yeah awesome, I'm desperate for that by the way that would be totally else what you have to do when I'm debugging stuff is just edit the pipeline and do a sleep, and then you can ssh. You know well keep ctl exactly inside, but having the terminal. Viewing octane would just be totally awesome.

A

Yeah yeah super.

B

Slick, it would be really really cool. I mean.

A

Life is a way of passing in the shell, maybe.

B

A

That's what I was doing or query.

B

Argument or something or.

A

Yeah, it would be.

B

A

It'd be really nice if you could just from right here click. I I really like what you were talking about of like if there was a button there- and you just said you know, give me this and essentially exec this and then drop me into a terminal.

B

Yeah, exactly like a terminal link here or something yeah.

A

B

The way this ux isn't awesome, um maybe a drop down thingy would be good or something because really, this is the breadcrumb, and these links are kind of tacked on the end. It should maybe be a drop down on the right, maybe your buttons or something I'm not sure about.

B

But I mean that's a slight ux fail, um but it was just a a quick way of getting those links on the page, but I turned the link here which could link to essentially the terminal tab on the pod page like this one with optionally, specifying the the bash or shirt or whatever yeah.

A

Whatever the entry point is right,.

B

That would be just amazing, I'd be desperate for that for the longest time like an easy way to open terminal in the pipeline. That would be.

D

I wonder if you could maybe have like a check if there's a magic, annotation or something on on the part to say that describes the shell or something that's a.

B

Great idea, yeah, we could annotate the pod and say the shell. Is this and then just get optin to take it from the ship? The port yeah.

A

Yeah, I think we'd be open yeah to supporting that.

E

At some point, we're going to add the preferences too, so this might be a good candidate to to be stored in like a global preferences, and so you can change it. Yeah from the some kind of preferences ui.

B

That sounds awesome. Oh just another thing by the way about this whole ui right now, this was just like the quickest thing. We could do just to get something really useful that the jig is x. Community can kind of use the ideas behind this ui could work with completely vanilla tecton. So if you're not using jinx at all and you're just using techtone, it should be pretty easy to patch this ui, so it would work with vanilla tecton. The implementation right now is slightly cheating to visualize pipelines.

B

It's using there's a custom resource in jenkins, excellent, called a pipeline activity that visualizes this kind of flow um we should be able to. If I show you permanent activity there is. um These are the pipeline activities, which is the thing that actually implements the visualization of the pipeline. I'm sure you said this one and then view the yaml. um So this is the actual crd that the uoi is based off.

B

We should be able to we use almost exactly the same code and use vanilla, tecton, pipeline tasks and task runs and those kind of things. So I'm hoping to make this completely generic to techton as well as genderjx.

B

It's just the quickest thing to do was just make the jpg version, but we should be able to handle either completely vanilla, tecton or jenkins, x and detector, um because I noticed there is a tecton plug-in right now, it's fairly simplistic and it's mostly for visualizing metadata, protecting.

C

B

But I'm hoping we can kind of build like a reusable pipeline view that works for both, because it's the same technology. We just slightly cheat with this crd. It gives us that kind of an easier way of visualizing what's going on, but we could easily generate this in the plug-in rather than having a crd.

B

So we can use the same kind of view if you like whether it's vanilla, tecton or not,.

A

Yeah, uh jason hall is the one who did that initial prototype of the tech time plug-in and I'm sure he'd be interested in and uh getting like, like you know, having this work with just generic tecton uh customer.

B

A

B

Yeah um another thing I found amazing by the way we've checked x, uses lots of uh jobs like cron jobs, garbage collection, jobs and all this kind of stuff. This is still pretty early days, but we've got this little operations view that just shows kind of jobs. I think about you, but I've always found cron jobs in kubernetes, really kind of really unique and how to understand whatever I've been able to just re-trigger jobs from octane is so super nice.

B

If a job fails, you can just retry it and look at the logs of all the jobs and whatnot, it's so hard, so easy to miss that a job is failing or whatever it is. um So we found that even just the current jobs ui in in august to be really really useful, but we're hoping to make a real, dedicated ui to help people install your physics from scratch view the insertion happen.

B

You know iteratively and then diagnose when one of the many different things in such a physics goes wrong, like an upgrade job or a garbage collection thing, or if your cluster's getting you know out of resources or whatever and there's a bunch of other issues that I'm hoping to dive into soon.

B

That's not really anything to do with junkies x, but it's just stuff we use in jenkins x. um So we use something called cuba, healthy um and external secrets. Excellent secrets and cooper. Healthy external sequence is an awesome plug-in if you've never used it before it's a way of managing secrets, but storing them in something like vault or amazon, secret manager, google, secret manager uh as your secret manager any of the secret managers.

B

So it's a way of dealing with uh secret stores, um which for git ups, is a godsend, because you have to deal with like seal secrets which are kind of a bit hickey to work with. So these are brilliant, but we could just do with a visualization of the crd that these are based on there's a cid called an external secret which shows and using secrets and external secrets. We could visualize what secrets are populated, what are valid secrets, what need entry?

B

We can even have a ui for entering sequence to you know, update a password or a token or whatever, through the ui, so I'm hoping for an external secrets plug-in very soon and cuba healthy is another one.

B

um Cuba healthy is kind of like a generic kubernetes health check tool and again it makes crds that visualize health checks, which are extensible and anybody can build a health check, and it would be really easy to make a really lovely health check page that just visualizes cuba, healthy health checks, um but shows them in kind of a nice way. So you can see lots of nice, green ticks and something's red. You can zoom in I mean sort by. You know red and whatnot, maybe some donuts um so yeah.

B

Those plugins would be super super awesome as well. So I'm hoping we can gradually grow the the auction community and get more more people keeping his econ system, hacking them again. Speakers yeah! It's.

C

B

Fun: it's it's! It's so fun! Writing plugins! Because they're easy you get a quick return, you're not lost in webpack hell or any weird stuff like that. It's really easy! uh So yeah! I love what it's awesome.

B

A

Excellent yeah, thank you for the demo. This is awesome um yeah. I I this also makes me think of some of the longer term ideas we have around allowing plugins to expose that what they so like say. There was a cooper, healthy plug-in, allowing other plugins to know that there is that plugin is also installed and then being able to say. I want.

B

A

To use its view in my plugin.

B

Yeah a link to it like I know I can link to it because I know it's.

E

B

Yeah and that kind of stuff yeah there was everything.

D

I think I mentioned.

B

Yeah I mean, but you know it's already awesome. Well I mean we could just keep gradually improving it in a few ways. One thing I'd like the idea of is, I think I may be rambled about this on slack recently.

B

I'd like it, if, like this nav bar, gets huge, pretty quick and having a switch, and it's now for everything all the the numbers. Great, I can switch to anything there. That's awesome, it'd be nice to almost have a little bread, crummy thing here where, if I've got you know, each plugin could have five views each. If I just want to switch between the five views in one plugin have a drop down or something or something here, yeah. So.

D

B

Do switch navigation, quick navigations between a subset of views, you know and then there maybe different plugins could link to other plugins, so related things could be together rather than having this big thing over to I mean I love this and I love the fact that it's expandable and it's meaningful and it's lovely, but it it soon gets quite big at some point we might want a way of I don't know, maybe having different context or something that you can maybe show subsets of this or something or maybe a plug-in, could show a subset of this, and then we could expand back to the whole thing or either something but yeah.

B

E

Would horizontal tub work for that instead of breadcrumbs.

B

um I've I found um I get well. I guess navigation is hard right in general. um Let me show you the example. I might be looking at this pipeline and then from here. I want to look at a pod and then now I'm looking at a pod, but now the pod is in the workloads place, but now we've kind of lost the breadcrumbs of where I was. I.

E

B

You know what I mean, it's kind of I mean I'm I'm talking like nirvana here, looking like the perfect octane. Eventually, it would be awesome if a pod, if a plugin could embed bits of the classic octane views, but edit its own kind of navigation here a little bit. You know something like taking over the breadcrumbs or something like that, so that we can navigate into these places, but then we can kind of go back again um but yeah. I should maybe use more tabs in the checkers x plugin.

B

I should try that out um navigation, that's tricky. I think.

C

B

Downs would be a great if we just had a slightly richer breadcrumb thing. I think with that would go a long way to making things better, but yeah we can.

A

Play this is actually really interesting, so I hadn't actually thought about the desire to want to take like a summary view and then have it be under your own navigation and a plug-in, but it makes total sense, especially when you look at it in the context of something like the jenkins x, plugin here, where your your pods are part of this. um You know it's part of this.

A

This build step and- and these all these pods are uh a step in that that process and you might just want to dive into one in more detail, navigating away from the the plug-in context into the uh default octant context. Kind of breaks that I guess immersion for lack of a better word like you're you're. No.

C

A

Plug-In experience you're now back in the octane experience, and I and honestly I don't, and someone can can someone else on the call can yell at me. If I'm crazy, I don't actually think it would be that hard to do because of the way we render our content. I think we could probably expose a plug-in endpoint. That was something along the lines of uh you know like an embed embed content response where you pass us the path and we send back the the the the view.

A

Like the you know, the the view, and then you nest that view. In your view, components yeah right, because because we all of this stuff just gets rendered and fired back as json and so and and right now what happens internally in octane, you set the content path and then our our path matchers match that to you know it says: oh okay, well, it's the namespace jx, it's a work! It's it's pods! It's this and it has this name.

A

So we're going to generate the summary view for that pod and the api could could take that same type of path. So, in the same way that you've created a link to it, you would pass it the path and then we would just return the view, but you would have to then pack that into your view, component list.

E

It's like iframe, basically right, and so um I mean not from an implementation point of view, just imagine having an iframe where you can specify. I want to see my part details inside this frame.

A

And then, because of the way we render the view where we just loop through and we render anything that's in the json stream. uh In theory, it would just work yeah.

B

We could we could try it and see what happens. I was pondering about the symbol of the um like, so I love the pods view and the deployment view, but in a lot of classes, I look at it's kind of big like if I look at pods, um often if you're looking at pods, you get. You have a lot um like a lot of pods like I have I've only got 13. This is quite a small cluster.

B

I've only got 13 here, but fairly soon, like if you've got 20 pipelines running at the same time, you have huge amounts of pods and there's different ways. You could kind of group things to then navigate to pods, so I could imagine uh like there's the apps crd, um that's starting to be used where you can group deployments and resources together by app. So I could imagine one day having an app crd plugin.

B

That shows you like the high level apps like I've got three apps and one of the apps might have two deployments and those two deployments might have three replicas yeah yeah. You could almost do a tree kind of way of grouping your pods. It's.

C

Still using the pods views.

B

But it's kind of yeah putting them in a slightly different tree.

A

Yep yeah and I think that's what we're we're slowly working on that, like with the the top link there. The workload viewer.

C

A

Nav bar, which shows the you know the donut graph and like it's just like a snapshot of the application. That's eventually where we want that to go is, is you know, you click there, and this is like a you know. You can have thousands of pods, but what you're seeing is just you know the uh the individual applications themselves, and then you click on that, and that takes you to more like the resource viewer instead of the instead of all your individual pod views.

A

um The issue that we have now is that this right now is just based solely off of uh owner, refs and labels, and so, like you can see here, it just shows all the pods individually. It doesn't like group them into uh into logical, um like components.

B

Yeah jenkins x manages to mess up this view a little bit because it well these tecton, really all the tecton pods look like an app which is slightly over the top really because each one's totally stuck, but uh so it'd be nice to maybe figure out a way of grouping them just in a big pipelines app or something like that or whatever yeah.

B

I know on the jenkins x perspective exactly we have a company called lighthouse that has kind of one two, three four five, five bits that are different deployments and things and controllers and whatnot it'd be kind of nice to have slightly higher groupings. So we could group deployments together into a higher level construct, but yeah it's it's already awesome, but all these things would be even more awesome.

A

A

Well yeah, this has been excellent. Does anyone else have any questions or comments before we move on.

E

I just want to say this is a totally awesome. I think this is my favorite plugin at this point.

B

E

You win good job.

B

Awesome beverages on me when, when we meet in real world.

A

A

All right well, thank you yeah. This is again.

C

It's been great, really appreciate.

A

It and uh I'm gonna, um I'm gonna. I was telling folks earlier we're doing a new thing on the octan.dev website, where we're gonna have a featured featured plugin, so you'll be the first post for featured plugins and uh I'm gonna use that if it's okay, I'm gonna use that video that you sent me earlier and.

C

A

That, with the blog post as well, perfect.

B

A

All right: well, we will share my screen and we'll move on to release talks. Let's see which one is it. Is it this? No, it's not that one go.

E

A

Is okay, um so we are going to start prepping for 0.15. uh This is a smaller release. We're we're we're trying to get back into. uh uh You know committing to that kind of two-week cadence of putting out releases. um This one has a couple of useful fixes. One is the context selector right now it puts an ellipse there, but it doesn't it just it just keeps the prefix and so like, for example, all of my prefixes look the same. It just says www.dot, and I have no idea what the actual cluster is.

A

So sam did work here to fix that um it now puts the ellipses in like the middle, so you can get some good context and actually makes you able to switch between clusters. um The multi-cube config bug that was reported.

A

That was a regression that's going to get fixed in this release, and then there was some work from scott andrews. He added the web hooks uh view which was under this api server menu. We've we've renamed that to just the web hooks menu, but the api server links that you will see in the resource viewer as a result of this will still work. um We just the we, the api server as a menu entry wasn't like super useful.

A

uh It was like too much of the low level detail being surfaced up, so we we put under web hooks for now. We want to address the issue of the menu bar potentially becoming long and like pushing things down and making it hard to navigate. We don't have a good pattern for that. Quite yet.

A

We're still thinking about that, but uh we know it's something that uh we'll want to take care of in the future, especially as we get more plug-ins and and if you had 10 plug-ins installed in octane right now, you're doing a lot of uh a lot of vertical scrolling to try and get to things so um yeah, that's just something we know about, and then uh the stepper component, which was recently merged in, is a clarity component that lets you kind of build out a a form wizard.

A

You can. uh You can do uh requirements like validations on the fields uh it uses our standard set of of clarity, form fields um and just packed into uh json that gets rendered out into steps. uh That's supported both on the uh for both internally as well as plugins. There's a there's, a unmarshall edition, so plugin authors will be able to guide users through steps and then process those forms. The the stepper component itself has an action handler that you assign to it and then you would for plugins.

A

You just handle the form in your action handler um and did I miss anything in this upcoming release? Sam hold on.

A

Scott cool all right, and so then uh moving on to what's next so we're going to be doing some planning tomorrow and uh part of that is, there are there's a there's, a cloud there's a whole host of bugs that are a little annoying right.

A

Now, um there's some ui jitterbugs, where jitterbug um there's some ui uh bugs where, like we, we put in this loading menu, which is helpful, uh it indicates that there's stuff going on, but sometimes it can, it can flash the old screen before the new one gets presented and when it does that, if the new one is presented quickly, you barely notice it.

A

But if it's like a really big resource that takes a while to load, you can be on that old view for like two or three seconds, and it can feel like it went back and like nothing is happening so because the loader disappears.

A

So um we talked about it and I think we have some there's a couple options: we're going to explore for how to address it, uh both on the server side and the front end side of things. So we're gonna, that's one of the big ones. We want to get fixed. I also want to give some attention to um operating and restricted like our back restricted clusters.

A

I really want to uh get us so that we're kind of degrading, gracefully and I think the the solution that that we've ended up at is- is right.

A

Now we're trying to be really smart and like oh, we'll, retry and we'll back off and then we'll retry in case the credential changes and you get access and all this I think what we'll end up doing for the first pass, just to get people a much better experience is if we try and we get unauthorized, we just cross that resource off the list and we don't try to access it again until they explicitly refresh their context or switch switch to a new one right. It's like your con.

A

Your thing expired when it refreshes we'll try again, but we're not going to try to do all of this uh incremental back off and like guessing and figuring things out, uh we're just going to you know, try it get unauthorized cache that we got it unauthorized and then return empty for that uh type until they retry and we're going to tie that into some of the work.

A

That sam did, because with the combination of being able to upload a keep config and watch for the cube, config changes, which was already there we're able to now, we can actually alert a user uh which we don't do now, but we can alert a user that that their cube config has changed and and and suggests that they switch contexts or reload um right now. I think you have to fully like replace the file to get that to happen if you could.

A

But if you just like stand up a new cluster and it injects a new context, we show that context in the menu, but we don't refresh the current one. um So there's there's ways that we can. We can watch for the file to change now because, because sam has code in there, that's watching for that file to be created.

A

uh We can leverage that same, uh maybe leverage that same code or refactor it so that we can so that if you're looking at a cluster and your access goes away and so we'll alert you that you tried to view pods and you it's not accessible. And then, when your helper puts in a new cert, we can say: hey your your configs, updated click here to refresh it.

A

Maybe we can do it automatically, but for now it might just be a manual as long as it's as long as it's telling the user exactly what's happening, um the other set of bugs that we have are mostly around just like little little oddities that I want to make sure we're actually digging into and exploring cpu spikes for for certain views and and just resource usage and uh general kind of slowness and things like issues that are a little hard to kind of dissect and take some actual investigation.

A

um A lot of those we don't get to when we're working on other features and smaller bugs because we don't want to take time away from those larger features, so we're going to just put an investment into kind of knocking out some of these bugs our previously to to these last releases, our bug volume, our issue, volume kind of grew at the same rate that we were burning it down and then lately our bugs have kind of gone up uh faster than we've been burning them down, but we're still adding features.

A

So it's time for us to just kind of do a do, a bug, a bug, triage release. So that's what you can expect after we get the dot 15 release.

A

D

Can I ask a quick question.

A

Sure yeah, that's the part of the meeting, we're at.

D

Now is open q, a awesome uh you mentioned when you were just talking, then about alerting a user or notifying the user. Is that something that's? uh Maybe I should probably have looked at more of the dots, but uh is that something you have there at the moment?

D

I was just wondering because I'm just thinking from a pipeline perspective, it would be really nice to be able to say okay, something's, failed or even passed, maybe or something having some kind of notification things, but also when we get into like things like cuba, healthy views, you know, there's some kind of attention needed to be able to draw you to that. To that perspective, is that something that's possible now, when coming sam.

C

So at the moment it is currently not possible, or at least through plug-ins. uh I am working on it right now. It's uh it's actually coming along. I I don't know. I can't promise that it'll be squeezed into about 15. um yeah. It might have the code. Might I might have a pr made by the end of the week, but it may or may not be reviewed by then, um but it is certainly in the works and uh there is already an outstanding issue for it.

C

So multiple users have requested, for this particular feature: brilliant, fantastic.

D

A

Any other questions I'll, I guess, as I was going to say, uh if either of you want to create, it would be really helpful if you created issues for like, like the the drop down uh menu, you know ui and so just like features that you would like to have. uh Please create those uh so that we can, uh you know, get them, get them in our our backlog and and uh get them prioritized.

D

Kind of said it before the call before the recording but fantastic job. It's amazing, oxton james, has been telling me for years trying to get involved in some ui work and I've. Never I've just always been so horrified by getting into all the different web frameworks and everything and I had to go and actually put together a view myself within an hour, so it was kind of like it was like a alleluia moment. So.

A

D

Everybody that's been involved, that's great.

A

Thank you so much yeah. It's wonderful to hear.

B

So you need to make another view now. James.

D

This is true well well, I mean this is the great thing we can. We've got so many uh integrations with jenkins x. I mean james had two there. It's just once we get over the initial we're working on an alpha for uh jx3. Once we get past that it's basically just extensions and plugins, it's just creating a good experience uh and we've always struggled from the ui point of view. It has we've heavily used cli, but that's changing now it's we can just go. You know heavy on on the ui and it's fantastic thanks. Swagton.

A

All right: well, if there is no more, then uh that's it. Thank you, everyone, uh james and james. I really.

C

A

And uh yeah, thanks again and I'll see you everyone next week cheers.