VMware Octant Community Meetings, 18 Aug 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Octant Community Meeting - August 18th, 2021

Description

Octant community meeting is held weekly. We discuss and talk about the current state and future of Octant, demo upcoming features and releases, and preview new ideas we are considering for Octant.

Feel free to add any discussion topic or question you may have to the agenda: https://hackmd.io/CzaPxtmXT_SW8nEpdwvGzw?view

A

It seems like we're live. Okay, welcome everyone to the octane community meeting today is almost 18 and uh well we're glad to see you here. um Remember again that if you are not able to join the the live meeting like you know, I had some users reaching in the past saying hey. I cannot join amber that you can watch the recording on the youtube channel and we try to keep the agenda also up to date with the notes from each meeting, so you can catch.

A

It feel free to add your add-in, your name to the app in the list and uh without anything else. Let me share my screen here.

A

um There you go okay, so first status update has to do with um the plug-in experience survey.

A

Last week we published a survey that we tried to keep it short, so we can have feedback from all users in the community that in the past, have created plugins for octan or our interest in the vlogging authoring experience, which is one of the key assets, the key features of octan the ability to extend its use cases or its base features to cover some other kind of environment.

A

So um the uh the theme for this sprint and something we plan to keep improving in the future is how to uh make the whole um plug-in alternating experience even better, so your feedback is really appreciated. We released the the survey last week. We have had uh some responses so far uh from from all the options in the survey, which is basically one questions and four options, that's it.

A

So um what is the most important for users so far so? uh Well, the the commonplace between the two responses so far has been additional sample. Plugin language support. I mean we have a collection of sample plugins in golang type script and I believe, there's one in python, but you know the users so far have said that they they will really value having additional samples in some other languages and a more intuitive plug-in api, which I think it's a broader ask.

A

It can mean a lot of things. It can mean even documentation itself can lead to a more intuitive looking api, I'm not sure, but we're yet we're still collecting your feedback so feel free to add. Also your comments there. If you have additional requests uh for the plugin experience in often.

A

Right, okay, so um next item that 24 issues remember again, the the this sprint theme is improving uh the overall plugin experience and so far we have several issues in progress and some others in the review phase. So I saw, for example, the image manifest one that that was a great team effort. So I don't know if someone from the team would like to share something regarding that one or some other issue.

B

Yeah, I can say a few words about that. That was a long one. It took that's. Why get that run through all the checks and everything? It's still in the view, but I think it's gonna get much quickly quickly. uh We had a lot of issues because there's a lot of incompatibilities and different uh operating systems and things like that, but it should should be really interesting one and I'm excited that that's going to be in 0.24.

B

I think I demoed early version of that.

A

B

Couple weeks ago, so we'll probably show it again when we do the release really really really interesting new feature. That's gonna give us the ability to do totally different kind of troubleshooting.

A

Yeah absolutely yeah. It's again really helpful in troubleshooting scenarios, among others, so great work. There.

A

Yeah well yeah, so far good progress on the dot 24 release. um So that's that's good news.

A

Okay, there's no additional comment there we can move to. The next item is one that that's really interesting. I think it was some who wanted this. This item, the uh an updated kind plugin this one's in interesting because you know kind- is it's a really straightforward tool to have at least local kubernetes environment? So I don't know if you want to add there sam.

C

Yeah, do you mind if I share screen for a few moments, yeah for sure.

B

C

Okay, so kind plug-in this is fairly straightforward. um I haven't updated this since, like 0.19, and so uh this you can see. The version bump now goes from 0.92.23, because it's breaking and uh kind is somewhat inconsequential, but it just keeps the client go dependency synced, so we're just working off of the same uh cluster object, deprecations and a similar life cycle.

C

I think, as discussed in previous meetings, 0.2 is going to be the big one, so uh I'm kind of interested in seeing how other tools in the ecosystem adopts this, because uh that means a lot of the older object versions won't be uh registered on the server anymore. So uh I'm interested to see how that looks overall and uh that'll affect how often gets bumped and how other tooling gets bumped and cascading a bunch of things, but anyway.

C

The more interesting thing here um is that uh I'm going to start creating a bunch of issues for a lot of the features that are in kind, which a lot of people may or may not be using, but they fulfill a very specific use case like, for example, uh you could actually run kind in an air-gapped environment. You can load node images as tar balls and you can pretty much use pre-built, node images to do so, and so a fun idea here would just be.

C

Maybe if you worked at a bank or something and or maybe a as a government contractor and you need you- don't have access to the internet, you have to have some artifacts approved and you know saved on a drive somewhere and then you kind of just move it in somewhere offline. uh It could your work and you want kind to be working because you just because you know you might not be able to google this stuff or be able to pull down new images in real time.

C

um A good way to do it is to actually just have these tar balls and have octan have the plug-in have them all in the same environment, and you don't even have to remember these docs. You can just kind of have a ui to build your own node. If you need it, and so that'll allow users to play with different uh kubernetes versions.

C

So that's it there's also an interesting piece here, which is for upstream kubernetes development itself. You can actually use the kubernetes source code to build as well. Although I think, aside from people who are working on upstream, I see very few people trying to do this so anyway. That's uh where this is headed so far, so uh yeah. So that's it and then recap breaking changes so uh the pr we broke. uh We had some breaking changes, since 0.19 probably should figure out some way to uh cross reference.

C

Every time we break or like say you know, some kind of plug-in compatibility thing like. Oh, if you're, not working, if you're using octano.19 and if you're plug-in with spoke for 0.23 uh you're, just going to see an error component, but maybe it shouldn't run at all. I don't know.

A

Yeah, you know it reflect and thank you so much tim yeah it reflects on. You know, an idea that we were uh kind of discussing with cara. um I think was yesterday. You know how to create a mechanism to verify plugins like a often verified, batch or plugin. That says hey. This is the there's a plugin that that's known to work in the latest. uh I don't know often version or if there is some compatibility issues to be able to automate that validation or verification yeah. That will be really interesting.

C

Yeah- and I mean in some ways like by definition right- that's what semantic versioning is for uh that said, we are a little bit more aggressive in our breakages, uh simply because we're not 1.0- and I guess in some ways- that's why we're not 1.08, because we're still breaking the stuff. So who knows all right, cool uh I'll move along here? uh The next one is probably deprecation this one's kind of interesting. It's not so much an octet thing as it is so much a uh for anyone, who's watching and using protobufs.

C

uh This is uh kind of useful to know uh so, uh if you've seen how, if you've used this repo before for installing, like the I guess, the protobuf compiler it used to be able to compile using the go plugin. So in other words the runtime and the compiler were the same thing. But the folks at google realized that now because it's in the same binary, they have to update the compiler and the language runtime at the same time and uh and because protobuf compilers, they don't necessarily opt their language agnostic.

C

So they were having trouble with this because they support a bunch of languages like uh I don't even know where this is uh the developer guy protocol. Maybe it's this one, uh there's like a whole list. It's like yeah, c, plus plus c sharp dart, blah blah blah, so uh they broke it out. So now you have the protobuf compiler and the runtime that's separated, um and so essentially I have a pr up which is currently building. I think it's built yeah, I'm still checking looks good, um which essentially captures this work.

C

um I I've tried scouring github, I think, like there are only a few people who are doing this right now, but essentially you can see here that we've now deprecated this protobuf 1.52, because it's no longer supported it's code. Freeze and now we are using uh protochengo grppc, which is the language runtime for go, and now we just have a hard system dependency on the protobuf compiler, which won't matter anymore, uh we're just interested in this.

C

So, uh at the end of the day, it just simplifies the simplifies the pipeline and argo and mod a little bit, but this is useful to know so, if you're, using protobufs and you're still on the old compiler, you should probably move to v2 because they are, they have frozen the code, I believe- uh maybe maybe they haven't thing I saw come in here, yeah like most of the stuff six months ago, deprecate the module yeah, so they're not going to work on this anymore.

C

So if you so uh that's just uh for anyone who's watching uh just pay attention to this and the migration is not it's pretty. I would say it's non-trivial, but it's uh it's not awful.

C

uh Cool! That's it for me,.

A

Okay, interesting, thank you. Sam.

A

Any comments on this.

B

A

B

um Thanks for doing that, sam, that's pretty cool. I think it's good to. I think it's gonna change just to see the body work for um yeah.

C

Yeah one could argue- and this is also maybe a philosophical discussion, but it's more of like I think we spend a lot of time bumping stuff on github actions on things that may or may not necessarily matter right like one could also argue that, like our schema, is never going to change for protobus like the fact that we're using a frozen item doesn't actually matter because in some ways like we don't care anymore. But uh that's you know, that's a bigger discussion for how software development is done at large.

B

I'm a big proponent of you know being on the latest.

B

For many reasons you know just to avoid bugs and security issues, and things like that, so so I I think we need to do that and eventually we need to operate and I've been on a project, that's a little behind and always paid for that some way or another. So if we can maintain, I know it takes a lot of effort, but if you can maintain and stay on the you know that uh edge, I think it's gonna be really important for us.

A

Hey thanks so much yeah. It was great to to learn that okay next. uh Well, we I don't know if we already had discussion points and we simply called them status update. But now we go to open q a uh from me.

A

You know I was. I was uh going through the example plugins that we have in the repo, which is, I think, it's a it's a great starting point for uh potential plug-in authors out there, um especially checking which uh languages we already have there golang typescript and some content in python.

A

uh I had you know success mainly with the typescript uh sample um plugins, which uses uh both of these. um You know tools, let's say a plug-in library and the yeoman generator. It was great. You know really easy to get started. I think it's here. I I created a a plugin with the human generator just following the steps there and I was able to call it bug capture and the description says.

A

You know everything sample plugin maintained by me to do actually nothing right. It doesn't do anything, but it it was just you know if I can build at logging uh up to this point, I mean everyone can do it, but my main question was okay.

A

If, if I have uh you know this uh sample running, what will be the next step right, probably going to the components and see what I what I need to add to my specific plugin and add that code to the source folder, because I see that the human generator creates a specific file and folder structure and, for example, it generates a src folder with all the content or the code, but in general what will be a good next step in order to build a functional, plugin.

A

I don't know if it's specific for typescript or not, but.

B

That kind of depends what you want to achieve. I see you said in your description that your plugin is doing actually nothing. So I think you will be. I think, you're really done yeah no, but the next steps would be just you know to build the logic to build the um logic you need in your plugin, so build the screens by using components.

B

For example, you currently have one page, so you can add more more tanks and then build your plug-in logic, populate populated with data and do stuff like that, and then there's the whole other thing once you're done uh how to how to make that uh visible, and we do have a tag, your github tag that you can add to your project. So it's it's going to be discoverable as that plugin.

A

Okay, okay, okay, okay, got it yeah, because after that I I run some other samples there yeah the alert sample again typescript.

A

um Let me see here this one is really interesting because it it takes data that is here in the reference over here. um For example, the deployment data is the same one. It's a deployment called echo1 and well it's here, because if, if I'm a, I don't know if I want to create a plugin in order to provide, you know, introspection dashboard user interface to my tool, the resource viewer is a really great addition.

A

I mean this is really useful for for everyone out there who wants to build plugins that you know that it's it's not that complicated. It's there, the the um the capability to add a visual representation and relationship between objects to your specific plugin.

A

So this one for me it was really interesting.

A

Okay and yeah. I I tried to use the action, uh go length, action, um sample action, plugging sample, but now success so far with the golang sample, so I'll still need to to check if I'm missing something um but yeah, you know you know, take into account the feedback so far in the survey that most users want to see more examples here now I can understand why it's uh it's a really great starting point to create new plugins.

A

So that's that's great, so I don't know if there's any comment regarding this, that was basically my my main question.

A

A

Okay, okay! Well, I think that's it! For today we have a short agenda. uh If someone wants to add something comment, something that's great. If that's not the case.

D

I actually have a question. um It's about my my youtube story. I would like to know if somebody has an opinion about this. uh Can I share a screen, please yeah for sure.

D

Let me see if I can do this. Oh great, I don't have permissions to do it. I love you, I'm gonna have to post this. uh It's because uh I didn't not the permission system. um No, I will have to. I will post this.

A

You put the link in the chat. Probably I can share it. Oh okay uh yeah, so you need to add permission to the document.

D

Well, oh yeah! I can probably show you this.

D

D

D

My question is basically um liam: has some problems uh with how we lock information on plugins.

D

He told me it was kind of complicated to do it uh if you are using go, and basically I realized that the main issue is that.

D

If we go to go plugin client.gov, basically, if you don't send a json from the plugins to uh I mean the plugins can send a lot directly to uh the master directly to option on. uh But if you don't send this as a json, they will send this lock as a debug.

D

So uh yes, if you click in the one that says it turns out that uh yeah, if you're clicking that that one right there this is on the latest version.

D

So if you see that comment, it says if the output is not a json format, it would print directly to the box. This is actually not true in this case, but in the version that we have of claim dot go uh it does so you actually just go and do l dot, debug and print the line.

D

So my question is: if you go back to to the thank you, I have this solution. Basically, what I create is a logger using h, h, a h, c log um yeah with this. You can basically use what we are used to just use logger.info and it will show the log information in the in the console and any other place, but we are using our old version of this library. I don't know if I should update and create some kind of interface or if this solution is scanned. Alright, uh what I'm asking like?

D

I have the information there and it's on the main plugin, but I don't know uh what is the more clear way and somebody has opinion uh did it does this? Make any sense is he's kind of clear he's not.

D

A

Went on this idea.

C

Yes, so I guess I'm still having trouble understanding.

C

What is the problem you're trying to solve here.

D

The basic problem is that if you uh try to print anything using the log dot print f, it won't print. Unless you have a debug, you have minus you, have you are running often in a verbose mode, doesn't make any sense. Yeah.

C

Yep- and that sounds right uh yeah that would be expected.

D

But uh some uh liam and other developer was asking if we can allow the user to print use an info, so so they can see their logs yeah.

C

Now, if they get print, does that necessarily make it easier right like we want to think about, like as a plug-in author you're trying to print something, um and let's say you print something and it pops a bit into an event loop?

C

Are you just trying to grip it out of the output of the logs, or are you just interested in being able to pipe that specific section of the logs to somewhere that you want, in other words, like I'm, not convinced right away that enabling print actually solves the problem that they're having okay um or maybe it does.

C

Maybe it does I'd have to think about it. I don't know ignore me. I have no idea what I'm talking about.

D

Okay, uh yeah, I don't know yeah my question. There was pretty much to see if this solution was okay or maybe another solution, it was actually better like a daily client.go and I mean go plugin and just to make sure and create some kind of interface to hide the logic about from this but yeah. The main complaint is that the users want to be able to lock information without a verbose mode, because just kind of it was really hard to find information with using verbose mode.

D

Okay, so yeah. I can continue this on the on the channel uh on this slide, because.

A

Thank you, luis and sam great, okay. Anyone else any question, comment idea.

A

No okay well have a nice day. Thank you. So much.