VMware Octant Community Meetings, 10 Jun 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Octant Community Meeting - June 10, 2020

Description

June 10, 2020

Agenda

0.13.1 release
Fixed HPA printer bug
Fixed TestBed linter warnings
Looking at two issues that were reported after 0.13 release:
https://github.com/vmware-tanzu/octant/issues/977
https://github.com/vmware-tanzu/octant/issues/969
Storybook Demo
WIP https://github.com/vmware-tanzu/octant/issues/971
Startup without kubeconfig.
Loading cluster access at cluster client creation time.
CNCF Webinar on Octant happening June 17th at 4PM ET / 1PM PT, led by Wayne
Fix for cytoscape https://github.com/vmware-tanzu/octant/pull/982

A

Hi everyone and welcome to this week's episode of the octant community meeting today as June 10th 2020. We have a short agenda here today, but I want to go through a few things. That's happened since we last had a meeting two weeks ago. The last week's meeting was of course, canceled because we wanted to well not have a bunch of meetings that yeah yeah now I have a bunch of meetings now last week, hashtag black lights. So with that I'm gonna share my screen and let's get started.

B

Thanks Jonas yeah, so we are we're coming off the 13 release and we did a dot release last week to fix a couple issues. One was the HP 8 printer bug that Sam fixed up that that one, a lot of people were running into that pretty immediately. So we wanted to get that out quickly. I think we got that out the following day or same day even know, I think following day: wasn't it so yeah?

B

We got that out and then there was also just a little nagging issue which won't really impact people, but it it impacted us it's it's the testbed inject winter warnings they were spamming, stuff, making it hard to find real errors so that that also got fixed in the you know, not 13 one release. So we encourage everyone to upgrade and be on the latest release and, as always, please report any issues that you're running into you. There are actually a couple outstanding issues that got opened up.

B

I didn't put them here on the agenda, but they are. There was one that got opened up about certificates and connecting to octave was having trouble connecting to the sign self-signed cert inside a local network. That's something I've been looking into trying to figure out exactly what's going on there I'm having some issues with replication, I I understand, fundamentally, why the errors happening from like a like in the technology, but I don't understand how this person's encountering it. So I am keeping my eyes on that.

B

I'm gonna actually send a comment out on that issue to get some more information about what might be causing it in the setup, so I can replicate it and if we aren't able, if it's not something that acting can fix, hopefully it's something that we can at least put documentation around or point people in the right direction, and should they encounter an issue like this so and then the other one is a new issue that came in around contexts in context: permissions I'm playing around with cube CTX, which is a context, management tool and I'm.

B

Trying to replicate this. This error, where the context permissions, get get changed or mutated in some way, and then octant is no longer able to read that context and I'm, not sure if, if it's octants fault or if it's some other tools fault or what or how, what all permutations of this made. This thing happen, it's actually the it is. Where is it at octant crashes on man, I think it's that one yeah I need to change the title of that one.

B

Yeah, so this is the issue. We realized that it's it's getting a permission denied. This person was running octant previously with ODOT 12 and was not having any issues like this and then that cropped up so they've been helpful in responding with like their configuration settings and what tools they're using in the versions so trying to also kind of determine, what's happening with this one as well.

B

So if anyone watching this or hearing this and they've run into these similar issues, please feel free to to go comments on these on these issues and add some more data or information that could help us kind of resolve them so I will those are the two outstanding issues that I've seen come out of 13 that people were previously running on at 12 not encountering this issue, so those are the those are the main things I wanted to cover today, and then we have a demo.

B

So we talked a few weeks ago about some work that happened with creating some documentation around our components. This is useful for plugin authors as well as for us, because a lot of times we're digging through the source code to remember how a specific component works or what the llamo structure, Jason structure, or vice versa, is of that component. So we created some documentation through the work that that Sam led and then we decided that we'd be better to migrate, that to the storybook tool.

B

The migration is currently in progress. We've done most of these things. Sam and Malana have have taken care of most of these items and we are slowly working through the rest of them.

B

I I suspect that we'll have this have this done sometime next week, it doesn't seem like there's that many outstanding and some of us have a short week this week, with the hala company holiday being on Friday and such so likely it'll be sometime next week, we'll see all this land but Milan's going to give a quick demo of what we've done so far. It's going to give folks an idea of what they can expect when this gets released.

C

Sounds good Thank, You, Wayne I. Let me sure, hopefully the right screen.

C

All right, can you guys see my screen yep all right, yeah, as Maine said we, this is our second attempt of doing the documentation. It's really important for especially for plugging user. It's best for everybody else, Jerry we're trying to get it a little differently. This diamonds story book is really great too. That should help us doing that the proper way and as Wayne said this is the world.

C

Can progress I just want to show you where we are right now, and this may change slightly by the time we finish the work, but so this is basically the storybook page. As you can see, you can see the list of all the components and on the right side, you can see the cameras that shows that component and there's some now officer that you can play with for each component, so you can just go there. This is basically the data that's passed into component, so it's very useful to see that and try different options.

C

So, if I change that time step you'll see the updates the editor right away, some of the components we've done so far listed here. So here's the list some that I like to play- that is very simple text. You can just see.

C

You can see the text is updated, and this component supports the markups. If I change the stitch, it updates the component. So it's it's a really good tool, not only to see the components live, but to play with different options, and also it's good for experimenting. So some some of the early prototypes we done, for example, for the results we were done directly here in storybook, because it gives us developers really fast. The development cycle some other stuff too, and also the one important thing about the components is the talks.

C

So we story books has support for dogs too, and please ignore the white team here, we're still trying to sort that out and to make sure the team them it's the same as an ideally. This should be the same team as the rest of the application, but here we will give a little more information. You know how the cause that component used will. Should she live here as well and for each component to display the props.

C

Those are the component props that are passed into angular object and for each component we can display as much the arcs as we want. I think we still need to do a little more work and adding some more color here some description, but basically this is kind of the framework we're we are hoping to go through and it's it's really important for our plugin developers to have to like this.

C

So I got any questions.

B

Now this is really great. No one are we? Is there a? Is there a plan for.

B

Like how we want to structure what am I trying to say, like the the work-in-progress stuff, like resource based a new resource here and the preferences- and things like that is there is there- is there a plan for like when those things get upgraded or become official components? Will they then get moved into the upper components list, or will they stay down and others, but I guess that's what I'm trying to trying to figure out right, I think.

C

Once we are happy with something it should get upgraded one you know and become the official components note to the upper category thing is that this stuff at the bottom can be hidden. Okay, so I used storybook in my previous project and we had so many stories that were numbers shown to the user, but they were. There was a special key combination to show them inside the storybook, and we used those mostly for testing and, for you know, testing some edge cases.

C

Story book can be used in so many aspects, and one of them is Anton testing, because it's basically you have a page for each component that you can put in different parameters and then at a neo site person or some other tests against it. Yeah.

B

Yeah I like that, okay I, wonder if it would be possible for us to actually have a toggle in story book that says, like show experimental, so that way the default view when we deploy this the default view of story book is only show things that are other core features and then they could toggle experimental and see the the other things is. Do you know if something like that's possible? Oh yeah, yeah.

C

You should be able to do it. Some additional controls inside the story book like but changing. The theme should be that hard, and it says it's sometimes really useful to see how the opponents behave in different themes and showing the more experimental features and stuff later.

C

Any other questions comments.

B

Sorry thanks. Thank you thank you and then I wanted to call out two other pieces, so we're actively working on on things, and some of them are larger things. That kind of dissect big pieces of the codebase and I wanted to call those out just because they're there there hasn't been a whole lot of movement recently, because some of these more larger back-end changes are currently being worked on, but Samms is currently focused on removing the hard dependency on needing acute config to start up octant.

B

This is something that has been asked for for a long time. It also is something that kind of it leads the way into some of the refactoring. That needs to be done to allow a stop rate as a desktop application, but also in cluster deployment. So I don't know if Sam wants to add anything to that or not, but I just wanted to call it that that's currently being worked on.

D

Yeah and I'm trying to do my best to break it apart into smaller issues, and one of the reasons that I haven't been creating big issues, for it is one I'm, not a hundred percent certain about direction to be taken with it, but also there is a.

D

We just have a lot of internal code that probably should have been split up a bit more to begin with, and so a lot of this might just end up being like shuffling code around first in a way that makes more sense for the reef actors to happen first, so once that is done, I think it can. We can start making this into better bite-sized chunks of work, and if people wanted to pick up issues on the way we can or if not, we could just do it all in a big push.

B

Great yeah, thank you, Sam.

B

The other thing that I wanted to mention was this is work that I've been doing. That's been going on for a little bit now, but it was loading cluster access at cluster client creation time.

B

So what that means is right now when Aachen starts up, it makes some assumptions about the resources that it has access to in a cluster and given those assumptions whenever it tries to list or get or watch those resources, it'll make an API call out to the kubernetes server and then get back whether or not it has permission to do that thing, and then it kind of stores that permission in a cache and attempts to refresh that during every attempt to list or get that resource.

B

If it fails, if it's, if it's unable, if it's told by the API server that it's not allowed to get that resource, then it goes into a back off state where it will attempt to refresh that permission over time. But it's still doing it when the list or get calls are probably are, are being accessed.

B

So we've created a new library that will go and and essentially take your config and it will go to the subject, access API and it will pull in an entire list of everything that you are allowed to do in the cluster, and it will map all of that into this cache of the resource and the verb.

B

And then we do that at client creation, time versus at startup time or at the initial load of octant, and what this allows us to do is instead of having to now check whether you have git or list or watch permissions during every single API call. We can make a safe assumption that you have those permissions, because you started octant with a config that did have those permissions.

B

The other benefit is that, should that permission go away, so we go ahead and we try to do the list call and it returns unauthorized from the API server we can now.

B

Then we can invalidate the entire cluster access model that we have and then force it to refresh itself, and so in the case of short-lived client certificates, where your access might go away and come back intermittently, as as you wait for some third-party tool to refresh that cert or or lay down a new cute config in your directory, octant will no longer be attempting to do lists and get some resources.

B

That'll be waiting for the actual cluster access to come back and then it will get all of the there'll, be a explicit, refresh command that the user can perform. To this say: okay, I have access again go ahead and refresh or octant will say: hey I see that there is a new cube config. Do you want to refresh your access to the cluster, and then you can just click? Yes, and it will do that.

B

So it's it's creating a kind of a new way to for octant to not only interact with the cluster, but also inform the user that that something has gone wrong with their current access right now, octant will just continue to to spam an API endpoint and to not really be informative to the user.

B

Just things will kind of just stop showing up and it'll dump an error messages out into the console and all these things, so the work that Sam is doing to allow octant to load without a cute config, combined with the work that I'm working on for cluster access will kind of remove a whole host of errors that octant currently suffers from, as well as make the overall user experience a lot better.

B

So that's it there any questions on that.

B

And then the final thing that I see Jonas added in there is June 17th 4:00 p.m. Eastern 1:00 p.m. Pacific time we're doing a CNC F webinar on octant.

B

This is this will be a guided kind of exploration, of a creating a normal like containerized workload and migrating and moving that into a cluster and some of the issues and and things that kind of happen and come up while you're doing that and we'll be using octant, as well as other tools to kind of navigate that and and figure out kind of explore the different things that can go wrong and how to fix them and and how octant is working to kind of give you all of that information in a way.

B

That is more easy to understand, then running a bunch of CLI commands.

A

That sounds awesome, yeah, we're this is a part of the CNC F webinar series. They have webinars on a bunch of really amazing cloud native, focused projects and being one of course, so yeah this is gonna be. This is gonna, be super fun and yeah. Please just click the link and register. So you have it on your accountant.

B

Right anyone else have any questions or anything ahead.

D

I can add one more: there was a fix to the site escape component, which is used in the resource viewer, and all that really does is, if you kind of notice right now in 13.1, it creates a stylesheet to style. The resource viewer click on it, but it doesn't destroy that if you have to get away from it so for people who have often running perpetually for a long time and is weaving in and out of resource beavers, it might be worthwhile to use the nightly build at the moment.

D

If you want to squeeze it a little extra performance.

B

Awesome I will add that to the agenda as well.

A

Awesome anything else anyone wants to add to the community meeting for today.

A

Alright, with that finishing up the finishing up there, the community meaning for this week have a fantastic rest of the week. Everyone and see you all next week have a good one.