VMware Carvel Community Meetings, 28 Jun 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Carvel Community Meeting - June 28, 2021

Description

Carvel Community Meeting - June 28, 2021
We meet every Monday at 11:30am PT / 2:30pm ET. We'd love for you to join us live!

Team discussed a few backlogged items in ytt and imgpkg. Full agenda here:
https://hackmd.io/F7g3RT2hR3OcIh-Iznk2hw#June-28-2021-Agenda

A

Hi everyone welcome to this week's edition of the carville community meeting. Today's date is june, 28 2021 um just a reminder. These are being recorded and we do upload them to our youtube playlist.

A

We ask that everyone who does attend these meetings to please read and abide by our code of conduct, which is listed here at the top of the agenda.

A

If you're joining us from home and you're you're watching the recorded version of this, uh we do invite you to attend these live. We meet every monday at 11, 30 a.m, pacific time, and we also do offer our karbal office hours, which we meet every second and fourth thursday of the month. Also at 11 30 a.m. Pacific time and the office hours are specific to those.

A

If you have some help that you need from the team regarding carvel or if there's any kind of more in-depth discussions that you want to bring to the team or provide lengthy feedback for for something regarding carnival, that's a perfect opportunity for for you to present at that uh particular meeting.

A

Also, when you're attending the asset you put in your name and any organization that you're representing, so that we can keep track of the attendees that that do come here live and we can keep the lines of communication open and um hopefully continue reaching out to you um after the meeting ends.

A

If you have anything that you wish to discuss or have any triage help needed, you can put that down here at the bottom of the agenda and we'll get to that after we discuss the week's plans.

A

We have a couple of announcements here, uh just letting you know we won't be meeting next monday for the community meeting and we have a holiday here in the united states. It's observance of independence day, so the next time we'll meet will be the following monday for the community meeting and leave the office hours will also be that same week.

A

The one the following thursday after the week of july 5th.

A

And uh just iterating on what we've been discussing, the last few meetings, we've been uh part, we are hiring. We have three open roles. One is a software engineering role, it's a return ship.

A

What this means is that it's an opportunity for those that have kind of stepped away from working to care for a loved one and are now ready to return to work. So it's a little bit like an internship, but with a little bit more of a caveat to it.

A

Another option we have here is a software engineering role based in u.s, and then we also have one that is based in india. So if you're interested in any of these, please reach out to us on slack we're in the kubernetes slack workspace in the carville channel. If you don't have slack, we also are on twitter at carnival underscore dev.

A

So if you're interested in you have, um if you want any more information on those roles, please feel free to reach out to us.

A

We have two releases to discuss that happened last week. Does anyone want to go over this particular one on vendor.

B

um Eli is not here: he cut the release, but uh yeah kind of kind of. What's new, we implemented a few features um and a couple of them coming from prs. So thank you to joel holmes, osterman and jay bertazzi um for the prs, but uh yeah yeah, just just a few new features here.

A

Well, cool! Thank you. Next one we have is the image package released.

C

I can talk about this one um yeah, hopefully image package pool in version, 0, 12 0 is going to run much faster, which is cool and then there's also a couple of bug fixes. So you know like, if you did a copy, there was a chance. You would get a panic due to a race condition that that's been fixed and then you had an image log file with you know, duplicate image, references and you did a pull. Those duplicate image references would disappear, but that's no longer the case. They should stay.

C

So you can have multiple image references, even though they're duplicated, with different annotations. Persisting on a pool.

A

Okay, thanks dennis uh anybody else, have anything to say about either of those releases.

A

Okay, uh moving on to status updates, aaron you want to take over from here sure.

B

All right, let's see so status, updates uh schemas still very much in flight um nice. Does anyone on the team want to give the update for schemas over the past week.

D

uh Yeah, I can add a little detail here, um so this scheme is a feature we've had in flight for some time, so we've been doing usability testing to try to identify gaps there. As part of that, we are working on a couple of stories for error, message, improvements and documentation for adopting schemas.

D

I believe we also are looking for people to do usability testing. So, if you're interested in trying out the new schema feature, please reach out, you can get a usability test on there and then also just a note. That's still behind a feature flag, but we're planning to remove that sometime soon in the future.

B

Thanks gary um and then the next item is cap controller, the package manager apis. We are currently nearing the end of our code freeze period and finishing up on some of the uh bugs and and testing things that we wanted to stabilize a bit. It is planned to release this week. uh Tentative date is july, first thursday, um but yeah we're just planning to kind of keep keep moving along on the remaining items uh and then uh up next we'll be working on private registry authentication.

B

um Then image package uh continuing to work on some bug fixes and enhancements, um dennis called attention to some of the things that were released in the latest release. um We'll also be talking about some things uh in today's ipm or today's discussion of our backlog.

B

uh Is there anything else you want to call out from the past week no dot com? Is it yeah cool all right and then, as always, uh there is an area for discussion, topics or triage help. If there are any things that you would like to discuss in in today's meeting, we'll try to carve out some time at the end of this meeting so jumping over to the backlog, then I'll. First filter by the ytt schemas epic, and we can see we have a few things currently in flight.

B

There's the usability testing schema, v1 migration and populate missing info during schema error messages. Then we have a couple of things at the top of the backlog. um A couple bugs. I believe that we have discussed both of these bugs in a previous meeting.

B

um Would it be useful if we visited these again or are people comfortable with these ones? I'll, maybe pop this one open real.

B

B

Maybe give a thumbs up if you're, okay, to keep moving uh thumbs down, if you'd like to discuss all right cool, keep moving um and how about the same one? For this.

D

D

I don't think we talked about this one yet.

B

Let's see so unknown file location for nodes added via template, replace.

B

So, let's take a look.

B

B

So we can see that we're panicking, which is not great.

E

So we're panicking there on the output of the error message trying to convert the line number to a string. I think the position as a compact.

B

B

Gotcha yeah- that makes more sense for for this message. Then.

E

I have a question for carrie with the prs that you're working on now. Do you have any relation to giving line numbers to the template, replacement.

D

um Yes, yes, it's still work in progress, but I think currently um notes that are replaced into a template using template, replace don't have a position at all that includes line number, so I don't know it's possible. Maybe what I'm working on now would solve this, although I'm not exactly sure if I understand this problem exactly, but it's possible.

B

Yeah, it looks like john had posted a couple questions a while ago.

B

Part of the reason this may be in the backlog where it is is because uh we thought that that story that you currently are working on carrie would solve this. So maybe we should block this one on that story, but does anyone? I guess, have any questions um about this story or anything that we're gonna dig into more specifically.

B

Okay, let me let's see if that's 417.

B

All right so we'll add a blocker to that one then, um but yeah, that's another bug, so uh we'll kind of treat it as we. We generally treat our bugs we'll, I think, take a pass at it and check in after a day or so once we learn more cool following those two bugs.

B

We have some stories that we've discussed before removing the flag, the experimental flag, which also includes updating, docs and then a couple of docs focus stories.

B

um I think, that's probably more than enough for this week for the ytt track.

B

Any questions on ytt or schemas before moving on.

B

All right um so then, looking at the backlog again, I'm just going to skip past the ytt stories and the next top story is an image package bug.

B

um And this one has surfaced from some folks that are using nested bundles but I'll give y'all a moment to read through this.

B

B

So it's a bit of unexpected behavior, um let's scroll down a little bit and just kind of point out, so so garrett it looked like you were able to reproduce this um and then joao and dimitri actually had a bit of a discussion down below and even furthermore in in slack they talked.

B

I would say at this point they're getting into the implementation details like what a solution could look like um and as joao is pointing out as the tldr here is. We can consider having a field to provide additional metadata.

B

It could follow an annotation style.

B

I think at a high level, that's what it says.

B

um But do folks have any questions or want to dive into anything specific, or I can also pause on a certain bit of text if you want to spend more time on it,.

C

I didn't get a good amount of time to read like I haven't, read that thread between joao and dimitri, but manifest json were they referring to, like. I don't know what that was referring to. So I don't know how important that is to understand this bug, um but I think the bug reading it makes a lot of sense to me.

B

Yeah, okay, yeah. I think I think getting into the manifest json is, I don't know more so the implementation details at that point: yeah, okay, um but yeah yeah. Overall, it's it's a, I guess at a high level. This is a bug to uh ensure that our nested bundles are working properly um cool. But this is the top of the backlog for image package then does it makes if it any any other remaining questions before we move on from this.

B

B

Yeah so the ytt.

E

Discuss what was discussed in the slack channel that I saw on the vmware side of things.

B

um I'm not sure if there's a conversation in vmware side things this, this slack thread should go to kubernetes slack um but as for priority, uh we're still keeping, I guess, schemas as our top priority and then after schemas it would. It would essentially be the open api uh work on ytt, but we're we're just serializing on ytt right now and then after that is when we're starting to address these image package bugs, and there are two image package bugs and then there's also another k-build feature which we had discussed before.

B

But it's just to improve debugging visibility.

C

Yeah we want to go through these as fast as possible, but are there any is anyone blocked by this? Is this needed by cap controller, or is there another team, that's kind of blocked and waiting on this.

B

Yeah there is a team that is using um the copy functionality, and so this is uh blocking, I guess, their ideal path.

B

There are workarounds but they're less than ideal, so.

D

Yeah, that's similar to what I was wondering about like the impact of this. So if um this bug could cause the wrong shaw in the result, lock file, so if that is used later, it would produce. The incorrect bundle is that true, okay, that makes sense.

B

Yeah, so so, in order to I guess, to use like this workflow, um anyone trying to do this would be blocked, so anyone that is trying to copy.

B

With with lock output.

B

Cool good questions, good questions, I guess any anything else regarding, I don't know priorities or order of these things.

B

um Okay, then, I believe we have already talked about this bug before also, which is a plain image, should not fall back onto first image in the image index.

B

So that's when there's an image index, I believe we have some logic where, instead of returning an error or something like that, we instead return the first image in that image index, which is a little.

B

B

Cool, as do folks want to discuss anything further in this one.

C

It's really a description. We should just error out if they don't give us an image reference, that's not an image index, I'm not sure if the comments and the discussion leads to a different suggestion, but is that the takeaway.

B

um Yeah, I was gonna say like I, I don't know if this conversation had concluded necessarily okay. um So at that point I think it would still be up to the implementer. To kind of you know, propose a solution and and then take that attack.

C

Yeah erroring out, if you give an image index, if you don't give a specific image, that sounds like a possible solution.

E

B

um So I think really between you know these two uh schemas the image package work. That's probably still going to be plenty for this week. um Do we have any discussion topics we do not are there? Is there anything else that people would like to discuss.

B

F

I have a question, but it might be more suited for office hours.

B

Again feel free to ask, and uh we have some time so. uh The main question is whether we have the right people on the call.

F

uh My question was: I'm trying to I'm trying to call image package from a go program so basically doing an exact dot command, and you know just calling out to image package and I don't get any output from the you know, even though like if you run image package from the command line, you know it tells you like what files it's uploading and it gives you the the url and the the digest of uh you know the resource that it created up on uh on your registry.

F

You know when you, when I'm running it and go, I don't get any of that information coming back.

C

You're, building a like a command and the uh the command name is image package with arguments being the sub commands and flags, and are you attaching like a writer or something to the out and error like fields so.

F

I just yeah yeah, I'm basically well what is it the exec.command?

F

It has a combine, there's a combined output method that it returns from that. That gives you both a combination of both standard error and standard out and like it's completely empty like there's, nothing in it.

C

Yeah I'm wondering by default. What is it right to? I just don't remember: does it right to the is that, does it find the the sub process a tty, and does it write to that directly? You might need to basically give it explicitly.

C

You know in the out field and the error field, a writer like a g bytes, you know buffer and then, like just print that out, you should like. If you, if you're running into terminal and you get output, it should be possible to do.

A

C

You know the exact command, um I don't mind pairing with you after this call, if you like, um just to look through it yeah yeah, we did.

E

Something similar.

C

In our, I can also point you to the: um if you look at the image package and to end test, we do something very similar. We use exact to like print out and run command, run image package commands and we print out standard. Our standard error, yeah.

F

Okay, well, I will I'll hang on the call and look through that cool. Thank you.

B

Cool any other questions or topics, people chat about.

B

All right, I think we can wrap up this call and then you can leave it to the folks. I want to debug some stuff.

A

Cool all right uh thanks everyone for joining this week's edition of the carville community meeting. um Just a reminder: we do meet every monday at 11. 30 am pacific time, but we are not meeting next monday on july 5th because uh we are going to be observing the u.s holiday independence day. Here we will have office hours.

A

I think I misspoke earlier this uh earlier in the meeting we do have office hours set for uh next week um because it lands on the second thursday of the month and um we'll meet at 11 30 in pacific time, and that's just an opportunity for you to bring anything that you wish to discuss with the team that you're having help um need help with, or you have feedback on just anything regarding the tool suite we're here here to help you in that way.

A

Otherwise, you can find us on the kubernetes slack channel on cargo or on twitter, which is uh at carnival underscore dev, and we hope that you do join us live. We would love to hear from from you and how you're using karbal.

A

Thank you have a enjoyable.

A