VMware Carvel Community Meetings, 3 Mar 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Carvel Community Meeting - March 3, 2022

Description

Carvel Community Meeting - March 3, 2022
We meet every Thursday at 10:30am Pacific Time. We'd love for you to join us live!

Full notes here: https://hackmd.io/F7g3RT2hR3OcIh-Iznk2hw

A

Hi everyone welcome to this week's edition of the carnival community meeting. Today's date is march 3rd 2022.. If you are watching this recording from home, we encourage you to come. Join us live we meet every thursday at 10 30 a.m. Pacific time, if you have anything, you wish to discuss with the team whether it be you know providing feedback for something that we're working on.

A

You need help with using the tools anything like that we would love for you to join us live and we can work through it then- and you can add that discussion item down on the agenda where, in the section where it says, discussion, topics and we'll get to it towards the end of the meeting.

A

If you are unable to join us live there are other opportunities to interact with us. You can come find us in the kubernetes slack workspace within the carville channel or on twitter and carnival underscore dev and also all. As always, you can find us on github um if you have anything that you wish to comment on there provide feedback feel free to do so at any point that you are interacting with us or any members of the community in any channels that I've mentioned above. Please read and abide by our code of conduct.

A

And for those that are using the karmal tools, we would love to know more about your usage of of the tools.

A

We have this pinned issue here that you can provide those details in a comment to you can also add your logo if you're representing a company- and we can add that to our adopters page, it's just a way for us to know more about how others are using the tools, help us uh work on the different tools with with knowing how others are using it and improving in that way, and it also helps the community whenever they see different use cases on the value of carbon tools in in different ways that they might not have thought of before.

A

So we would highly encourage you and to drop a comment in there. We'd love to hear more about it, those that are on the call, please add your name and any company you're representing if you haven't already done so to the attendees list.

A

On to announcements, in case you missed it. On february 22nd we had a webinar with tonzu tv with one of our community members, scott rosenberg. He went over carbo packaging as a unified package manager for kubernetes.

A

You can click on that link and you can watch the video that we're recording and get some some more. uh I guess uh demoing about carnival and also he throws in some tce stuff, which is a tons of community edition, and you get to learn a little bit more about how to use our tools.

A

And this week we have a blog coming out, I'm not sure if it has it been published. Yet this link doesn't go anywhere.

A

See I know we were still reviewing it.

A

Yeah great, so uh we have a blog that we just posted today as a reminder, we're doing weekly content shares and this week's content is migrate, existing resources to a new cap app by praveen reward. He is a maintainer for cap, so please go check check this out, give it a read: work through it drop us a line of any feedback or or questions you might have it.

A

We would love to hear from you and if you you like this content- and you want to continue seeing more of it, we you know, we want to know that too, and and what other stuff you might want to hear about from the maintainers upcoming we have a blog next week or actually in two weeks. I believe this one's coming, multi-environment deployment with yttncap by yash satya and as a reminder, not you know, this content sharing isn't just reserved to our maintainers.

A

We would love for the community to participate if there's anything that you are working on. That includes carnival or anything. You find interesting. That includes carvel of anything related to it. We would love for you to share that with the community and it can be in any type of medium that you're most comfortable with a blog, a video, a workshop. You know whatever you you think, uh would be good to share that knowledge.

A

Please sign up. Add your name topic, the medium that you're you're going to be sharing that content in and a twitter handle. If you wish for us to use your your handle when we tweet it out.

A

If you have any questions about this, please reach out to us and if you do participate, we would love to send you a t-shirt as a thank you.

A

On to releases, we just have a couple of little minor ones here I don't know if uh someone wants to go over these.

B

Yeah, I could talk to him real, quick, um the 37.1 that ytt 0.37.1 is actually a back port. It's for our long-term support. um Folks who are using um our minor releases, uh 39 has the same fix um and then some so we'd encourage you to do that.

B

The 37.1 is only there for folks who have to stay on the 37 um version, so just a friendly back port and then the 0.40.1 that had to do with um learning how github does releases and they order things when we added the 37.1 github said: oh well, that's your latest release. Now like wait. No, no! We wanted 40 to be the latest release, uh so the cleanest we put our heads together the cleanest way. We could possibly think of doing this.

B

We've got a lot of machinery behind our releasing um was just create um this 40.1 release so that, if you come in and ask for the latest from github, you'll actually get the latest. So that's kind of all that was anyway, there's stuff brewing in the background, but we haven't released yet, but that's what those two are.

A

Silly github, I thought there was like a button. You could say no this. This is what we want. The latest release to be um it's a good workaround.

A

All right, moving on to status, updates anything to call out for the project roadmap.

C

um Maybe the only thing to add this week is we we did kind of have like a casual kickoff of the signing and verification um track. It's the last item in this road map, so just just kind of kicking off the research discovery piece of this.

C

We expect that this is a pretty pretty big and complex topic to to get into, and so that's why we do have a somewhat distant timeline for this, but uh just just mentioning that it's getting started.

A

Great thanks, erin.

A

Okay, so this week anything to call out for cap, I know that the teams aren't seems not present, but there's anything that they're working on this week. That should be called out.

A

Nope, okay, image package and cable.

D

Funniest thing just happened: I tried to click on my screen to unmute myself. um Okay, for image package, I've been working on, we've been working on the describe command. I think the scribe command should be pretty much ready to go the first version of it. It's a very simple version. Just provides you some yaml in text, outputs both of them, um and it could be helpful for um for you to check what the what the contents of a bundle uh effective vr.

D

So that's that's that has been ongoing and I think we'll be able to uh merge that today and create a release somewhere in the next couple of days. uh With this feature, and also the the prior feature that we talked about about freezing transports, so I think that's it for k build uh there's. There is nothing uh happened. I do believe that there is a bump of a just a dependency, and that was it.

A

Great. Thank you. Cap controller.

E

So I think, for cap controller, in some ways the the biggest news is um uh actually the cap team uh is getting ready to release a cli.

E

It's sort of like so the cap command line interface doesn't know about the packaging resource. The packaging api and the cap controller cli uh is is a way for you to interact a little more imperatively. um I know like when I played with it briefly like I thought the coolest feature that they had. Is this kick so that if you need to um right now, if you have a something, that's reconciling and it gets stuck for some reason and you you want to nudge it along.

E

You have to cancel it and then uncancel it or pause it and unpause it, and so they now like imperatively, you can interact with these packages by giving them a kick anyways it. It does other cute things, there's going to be release, notes and um uh anyways, it's gonna be great and I think otherwise, the caf controller team.

E

You know it's a bunch of um there's a list of small things. That's not immediately coming to mind, but uh they'll all be in our release. Notes.

C

And on that note, we'll likely cut a release next week for cap controller.

A

Awesome. Thank you anything going on with vendor this week.

C

No, I don't think so.

A

Secret controller.

C

No, although we'll have to take, take a look at those repos, but since we're cutting cab controller, maybe we'll cut releases there if there's some work. That's that's been done recently since, since the last release for this.

A

Okay, um okay, ytt.

F

um I'm working on making ytt portable to windows by changing the library key constant. This is just fixing a bug um uh set this down for a little bit while I was out, and so it's back in flight.

G

And I believe, last community meeting uh was the first day that we had kicked off the validations track for yt. um Today there is a pr, so we are making progress and super exciting explanation. Point.

A

Lots of exclamation points.

A

Okay, I have on here a demo carrie. Are you prepared yeah.

F

Yeah, I just saw you come in sorry.

A

Sorry, I saw scott come in, um go for.

C

A

Cool I'll stop showing my screen um carrie, so you can start.

F

F

Cool, so I expect this demo is going to take like maybe six minutes or so um I'm just gonna go through the blog post. That was posted february 23rd about um using a carnival ago, cd plugin.

F

So please interrupt me if you have like small questions along the way. um This is going to be pretty informal.

F

F

uh Argo cd was declarative, continuous delivery tool for kubernetes. um It uses the get op's philosophy of storing the desired state of the system in git.

F

In this blog post, the desired state of the system stored in git is stored in ytt templates.

F

So argo cd will template and deploy.

F

um You know your configuration, so it has built-in templating engines like helm, customize jsonnet case on it.

F

There may even be others, and then it also has like the ability to extend any of the templating engines with your own plugin. So people do this for um maybe need, like a specific version of helms, to be used or a completely different templating engine like well. What we're going to do using ytt.

F

F

I'm gonna walk the walk us through exactly how that works, so to create a plug-in with argo cv. You have to actually patch the argo cd cluster configuration. So one of these patches is the argo cd repo server.

F

Basically, we need to include the tt binary in the pod in order to be able to use it as a plug-in.

F

So what we're looking at here is um the configuration for the argo cd deployment for the rvcd repo server, and you can see there's some ytt directives in here. um This just controls exactly what uh configuration it's going to match on and then uh modify.

F

So to walk you through exactly what this does we're creating an empty directory and using this image that we publish in the vmware, tonzu, uh github, reposito, uh github organization, um with all of the karmal tools and we're gonna use an internet connect, init container to copy the ytt binary into that empty directory um and then mount it to the argo cd repo server here and then this ytt directive is just saying: if the uh array entry for the cd repo server already exists, we want to add to it rather than like, create a new entry.

F

Since um it's an array item- and there can be multiple of uh multiple array items with the same name, we want to specifically just add to one. If it already exists. There.

F

So this actually adds the binary um and then we'll have to do one other thing to just make the plugin like available to applications, I'm sort of exposing it to the cluster to any applications. You have deployed with argo cd, so that's patching the argus cd cm configmat.

F

So again, some ytc directives here matching on the config map called aggro cdcm.

F

If this data key is not here, that's okay, we'll add it, and the config management plugins is set up as a multi-line string. So it's actually an array entry, um we're going to add to it to add a new entry called carbo ytt, it's the name of the plugin and the generate step. This is the step that actually templates whatever is stored in your like source, git repository.

F

um So there's like a an initial step, there's a generate step and then, after that, it's the actual deployment to the cluster um so for generate we're going to just run ytt dash, f dash dot for all of the files.

F

um This command can be obviously edited. um You can add additional flags in here. However, uh it does get a little bit complicated, because this is somewhat uh like hard coded for the entire cluster.

F

For your argo cd configuration, there isn't an easy way to like dynamically pass in flags here say if you need to add um like a a data value from an environment variable, um it's not uh easy to do it using this argument here there is an open issue in argo cd that details how to do dynamic flags using environment variables.

F

um So you can use environment variables in here in those arguments, and you can also pass environment variables from your application.

F

So not covered here, but if you're interested there is an open issue and arguably for that.

F

So, let's deploy it.

F

Create the namespace for our cd and then we're going to actually.

F

Deploy uh argo cd along with those patches that we just saw so using cap to deploy an app which we're going to name argo into the namespace argo cd and we're using ytt to add these two overlays um that we just described above and actually let me show you so here. I have uh these overlays uh in a repository here, they're not in the website, but they're the exact same code.

F

So let me just show you. This is just like what we saw so when we do run this command, we're going to apply those overlays to version 2.25 of argus v.

F

And then the output that you see here is cap, so it's making us acknowledge the changes that will be applied to the cluster. So it's going to create all of these resources needed to install argo, cd, press, yes, and let that apply.

F

It's going to take a second.

F

um So now we actually need to create an application that we can use with the carbo plug-in right. So um the example configuration I'm using is um I have it here in my other terminal tab is the carville simple app on kubernetes, it's repository in the vmware tonzo organization.

F

um That has a bunch of examples on how to use various cargo tools and the one we'll be using is config step two.

F

So it's got two files, config and values um which I'll just show you what they look like, they're, pretty simple, so the values file that we have here is in ytt format. It's got this data values uh annotation at the top, and it's got hello message. Stranger.

F

And in our config file, we've got a service and a deployment, and the key here I just want to point out is we'll be using the value from our values file to substitute that in as an environment variable.

F

So this data value is hello. Stranger, and this is also the ytc template.

F

So when we apply these two files with ytt, we should expect that this environment variable that gets passed to our deployment will get substituted in here, as is the ytt fashion.

F

See how our that's working so um in order to create an argo cd application, we're going to create an application resource.

F

um This is a pretty simple resource. I added the sync policy to just like automatically deploy it um and it's pointing at that repo for a simple app on kubernetes, watching this specific directory for changes. If any commits are published here, then it will redeploy this application with those new files.

F

um Important part, add this plugin config with the name cover ytt um and that will allow you to use the carbo plug-in that carbo ytt name comes from the patch that we did to the config map here.

F

um There is a bit of a little work around here, so we're also creating an app project with name default. So this is just the default project, that's like automatically in the cluster, and we need this because we're going to be referencing it in our application.

F

um There's a bit of a delay before the default project is actually created in the cluster, so we're creating it manually here. So the demo is smooth, um but it shouldn't normally be necessary.

F

Cool looks like our deployment succeeded. So let's also apply this application to the cluster.

F

I'm going to call this argo app.

F

And it's going to create app project application.

F

All right, so that should be all we need. It should be templating and deploying that application to the cluster right now so, let's inspect it so first we need to just set the context, and then we can use argo cd cli to get the status of the app.

F

So looks like things are: synced and healthy got a service and a deployment here, um let's just view the app to see that it was deployed okay and that we do see the hello stranger environment variable was templated correctly into our deployment, so import forwarding here. So we can see the app at uh this address.

F

F

There we go hello. Stranger was used in the deployment for a simple app, so our templates were evaluated properly by ytt, using argo cd, and now we profit so pretty cool. That's all. Are there any questions.

B

Yeah carrie, you were talking about the app project in application dependency. It looked like it all just worked nicely. Was that just happenstance or um was there a weight rule that you had to add or yeah.

F

Oh yeah good question, so I kind of glossed over that um if we had excluded this app project resource um what we would have seen when getting the status of the app so this command here our go to the app get simple app.

F

um We would have seen that it was not synced and healthy right away um and it takes about a minute to actually become synced and healthy, because that's about how long it takes to create the the default project in the cluster, there seems to be um a bit of a delay using the um the default project um like right after creating the uh right after installing argo cd into the cluster, there's a slight delay before the default project is available.

F

um So my suggestion is create it manually for the purpose of this demo, so that we don't have to sit here and wait uh about a minute until this becomes synced and healthy, um but shouldn't be necessary in any use case outside of a demo.

F

Normally that that slight delay is not impactful.

B

Gotcha, thank you very clear.

A

Any other questions for carrie.

A

Okay, cool thanks carrie for showing us that, um if, if you're watching this from home- and you have any questions regarding the demo that you just saw, please feel free to reach out to us in the kubernetes workspace we'd love to hear from you now. Let me share my screen again.

A

Hey, we have a couple of discussion topics. uh John did you want me to take over as you go over this.

B

Yeah, actually, I could just talk to it real, quick. um I think the main point here is: I wanted to just talk about this on a community meeting. I think most of the maintainers already familiar. We've got this gone, um but um really quickly, so so added this little hack md to be a simple introduction into our ytt validations proposal in that hack, md, there's a link to a video shorts less than 15 minutes, and it's a if here's it's it's a gentle on-ramp onto like what would why? Why are we doing this?

B

What does it look like? What? What are the important parts in the proposal? If you wanted to just take a cursory look, where would you look because it's sort of a larger dock and then links into both the the rendered markdown, which is a much nicer way to navigate and uh the pr which is like one of the best places to drop feedback?

B

Folks have been doing that? It's been great, there's a lot of folks actually in this room that have been pushing the conversation forward. Look at that uh 43 comments and then I've had some other stuff uh folks coming in from slack. This has been great. We've been getting a lot of community support around this, which means that, as we um get the first version of this feature out, it's going to look much more like something that solves your problems in the way you'd like them to there's time.

B

Garrett's already was talking about how we already broke ground and we're getting started. The stuff that we're doing right now is not controversial. It's like plumbing that we need to have in place, but there's still lots of opportunity here to influence and shape the way that this thing ultimately looks like, but the clock is ticking so definitely wanna. I would love folks to take a look. uh You can drop your comments in our public channel uh uh in kubernetes, there's a link to it there or drop it on the pr.

B

um If you wanna dm me, I'm j tigger on that kubernetes slack whatever's going to be most convenient for you uh or you could show up here, and I would love to hear from you. That's all.

A

Great thanks, john, and um you said the clock is ticking to provide feedback. Is there any sort of estimated deadline, as far as when folks could add their comments like? What is there a a stopping point where, where it's just we, we aren't able to take anymore.

B

Not really like, uh probably practically speaking uh for there's probably a couple, um but there are some features that we're planning on that's a little bit more advanced that might come in in a second round. So there's some parts of the proposal. For example, the export to open api v3, including validation rules, is an example of that. um uh That might be a little bit longer.

B

You might have more time to kind of get in there and tell us about your use cases, um because we would love for, for example, for ytt schema to be the way that you get to describe your configuration api and be able to just export that to open api v3 for any tools that need that so, but that that's probably a little bit longer so yeah. Let's say a couple of weeks.

A

Well, thank you, and you know that's also not to say if there is something that you have to share about this, and it's been a few weeks, please do feel free to share it with us. We still want to hear about about what your thoughts are regarding this.

B

A

uh Okay, next step joe wow talk about human readable tags.

D

um Yeah, so so I just I, I decided this year to tell everybody that the contribution period has ended and that we're going to implement uh the the solution proposed there. We talked about that this last week to see if there was anyone that had any kind of concerns about the implementation, uh since nobody reached to us having any any concern.

D

uh The idea is that if you go a little bit up just to remind everybody, so there's there's gonna be a flag on your. um um This is going to be behind a a feature flag. So you, if you want to activate this, you can activate this behavior. If not, it just keeps the current behavior. So every time that image package copies an image or a bundle from one registry to another, it will create automatically tags for each in one of the images.

D

So this is a requirement because there's there are registries that require you to have it or else it doesn't. Don't even allow you to push an image that doesn't have a tag or there's other ones that if the images are not tagged, the garbage collector will start this process and clean these images.

D

So that's why image package is providing automatically tags, and previously the tag was basically just the sha of the image dot image package, uh as as you can see uh like on the example there, and the idea here is that now behind this uh flag and this feature flag, the tag will be a little bit different and it's going to contain part of the repository origin, repository name that will uh of the image.

D

So, if you see the example like the last example there like, for example, for the my registry, dot io slash my super interesting product, slash application, dash power, slash controller, the the name of the tag will be our interesting product application power controller dash, and we kept also the shadows there. So there's there's also a limit of characters that we can use, but we think that that could be enough. So uh this um we're gonna go forward with this uh change.

D

Alex already made uh some of uh the the groundwork for this and there's a pr there. So I'm just going to review that here and see if it matches with what we have here and if it does uh eventually we'll murder. The pr and this feature will be active, so just an fyi that this is going to happen.

A

Great thanks joelle any comments from the team regarding this anything they want to add nope. Okay, any thing that's come up any other discussion items before we part ways.

A

A

Well thanks everyone for joining this week's edition of the carnival community meeting. It's a reminder: we do meet every thursday at 10, 30 a.m, pacific time so come join. Us live if you're able to we'd love for you to love to meet. You love to hear how you're using the tools.

A

uh If there's any feedback you have regarding all items that we talked about in this meeting, we would love to hear from you. So please reach out on slack. Let us know your thoughts. uh We can't build great products without knowing how y'all are using it and the best ways that you're using it. So that, thank you and we hope to see you soon.