GitLab 12.5 Release Kickoff, 17 Oct 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: GitLab 12.5 Kickoff - Monitor:Health

Description

12.5 Kickoff for the Monitor Health Team

A

All right so now we're gonna do 12 toys all right kick off. This is totally informal, I think actually for this one, a lot of us hasn't seen the issues that Emilia and I have been working on. So what we have planned for 12/5 is not taking into account. What's gonna come over from 12 4 he's also need to be broken down to other issues.

A

We mean push them out to 12 6 so that we can create some spikes to dig into this new category or some technology that we're not familiar with so I'm just gonna start from the top we're gonna walk through these four issues and I'll have a melee I'd go through some mocks for a couple of them, so we're moving into arrow tracking error. Checking is based on an integration between century and gala.

A

Century is I believe the largest opens first era tracking tool in the market right now, they've been great in integrating and having a partnership with us over the last. However many months, so we're excited to continue that with them and in the next two milestones will be maturing era tracking the viable.

A

So, first and foremost,.

A

We want to enable people to get started on this platform really easily. This is gonna lead to wider adoption. This is gonna, make it much easier for people internally across skate lab to dog food this. So today we currently integrate with century. We don't offer people an easy way to deploy it. We want to add its.

A

We want to bundle it with gitlab so that people can deploy it to gate clusters if they have those and get running dastard, so stranding with a comment more about this this morning, and this would make it super easy for us to get the rest of kit lab dog floating this even after this first milestone. If we can add it to a few of our Auto DevOps projects, so proposal, add it to our cluster X list yeah and.

B

I have a question when you say cluster apps list what you're referring to visually right. It would be that that page of, like sort of one-click installs like where you can install prometheus, why not yeah.

A

B

We'll need to work closely with configure orchestration on this because they manage that part and I know also that they are in the midst of reworking the way that that works with, because they want to be able to have those configurations, be version controlled and not just be like.

B

You know you click and install, and if it's you have no history of it, so I guess all that is is to say: I will talk to some people there and make sure that they are aware that we want to do this and make sure that there aren't further blockers. Okay,.

A

And then I also wanted to point out. This is not concrete. It is just based on Kenny's brain. There might be a helmet chart out there for this, but we can talk to configure about that. Do you want me to tell you unless.

C

B

A

B

Also know that they were dealing with a few requests like this right now. There's a couple of teams.

A

B

To add new apps.

A

Okay, if it needs to be a bigger meeting, totally fine.

C

A

Moving on so actually, let's go look at for context, but the arrow tracking table looks like today. This only has one wonky error in it, because trying to set up a project with air driking is really hard.

A

So once this loads I can show you what the error list. Okay. Here we are so today we integrate with sentry and he pull in errors for a particular project or a particular application. That's running in a gate lab project and we surface it. Here we provide five pieces of information, title details of available events, users and last seen you can both go to the century dashboard for this project or access that error in sentry directly.

A

So the next couple iterations these next three issues are all about making that more functional, easier to use and giving you more information.

A

A

So the detailed view for centuries and collab, we want to display more information, we're trying to reduce the time people spend switching in between tools in an ideal world. We provide enough information in gitlab that allow someone to determine if it's critical enough to create an issue that is a visionary workflow I would love for us to work towards over the next two months.

A

Obviously we are not going to rebuild sentry areas and gilliam. That would be a waste of time, but if we can give them just enough information to make a decision, I think that's a great goal course goal post to go towards Amelia. Do you want to work through the designs that you have sure.

D

Sure so you can click on the default ListView.

D

So the idea for this issue would just need to remove the external link on the error title and instead of it going directly a century, it would open in a new page, which is the other screenshot and then show some of the key details about the error, and we will also have a link to the century detail.

D

Exception as well, but we'll try to at least surface some of the information we think it loves, so they don't necessarily have to go into century. I. Think what we talked about as kind of a first pass is the title. The description I have perhaps some error details, and then one thing we heard from users is that the stack trace is really important but I. Think probably that's gonna have to be a subsequent issue, so I guess what I'm saying is the details here are still.

D

We still need to discuss like how much of this we can include as a first pass and what we might have to pull out to be a subsequent issue.

A

You I assume, do you have any questions on this.

B

No questions I mean I, guess that if we are going through this thinking about you.

A

B

That will require more investigation. Certainly, this is one of those to figure out how okay, usually we can get those error details, or rather like how much work we would need to do to format them or use them and then also if there is an easy way to to like render the stack trace and in a useful way. Presumably, it's easy to get through the API.

A

So there's a couple ways we could do this from planning perspective um first and foremost, I think that we should commit to a little bit less and then we can make that investigation part of the issue or we can commit to a little bit less and create spike issues. I think we've been doing that I, don't know how, if everybody's happy with our format,.

B

Yeah and in this particular instance, I'd I'd prefer to just start with doing the smallest iteration and then hopefully, in doing that, we'll be able to learn like how easy it would be to do the details. Part for the stacktrace piece.

B

I think we could also just sort of leave that, as as an issue to be added on to add it on to this, and then we can revisit later whether we want some dedicated investigation into that or whether we've learned enough through what we've done. I.

E

Think I would advocate for a separate spike issue that way we can kind of work on it more parallely. It's just what comes spiked.

B

Issue, for which part that's.

E

That Crais.

A

Okay, awesome I'm, going to let you guys create the spikes and schedule them and add labels to them, because we're doing it on a case by case basis, so for the D for the issue we're looking at right now over the next week, Amelia and I will refine it down to the MVC.

A

Specifically, the stack trace is going to be something we need to investigate. We can add in 12.6 as an example, let's go on to the other couples that also have to do with this view.

A

So I don't think that we have mocks here, we're kind of all working off the detailed view mocks, because all of these features are within the same couple of screens, so we'll attach specific ones to the issues once we finish those, but so is to not have multiple discussions.

D

It's a mock up down below there's a thread below because there was some questions that were like specific to this issue. The mock ups need to be updated, but there are a few visuals out of there. Okay, speed.

A

All right, yay.

D

So this one would be primarily about adding that search or filter bar at the top. So this is just showing what you know. The basics of searching filter bar from like the issues page, would look like if it was added here and then on the next. The next PNG, so I think this is Clement, had a comments on this issue that we're going to need to break it down. According to the issue description, this is these are the things that we would potentially want to be filtering, release, requests or commit, as it sounds like.

D

Another team is also working on the release piece of it in in terms of issues. So I don't know if that's I, guess I'm, not sure which is the most important here and I'm, not sure. If maybe just focusing on one of these filtering items is a good first pass and I think that's one of the things that it would be good to get feedback on I.

A

Think that's a great point off the top of my head. If somebody else is working on release, I think that merge requests would be the most important one for us to focus on as that's how we would go and find a group of commits that caused that error of you looking feel free to disagree and I. Do think that we need to learn a little bit more about that. But that's just mine off the top of my head, ah yeah I was.

B

Gonna say I think technically: I I, don't I, don't know if an, but my hunch is that technically filtering by the merge request will be more difficult, because that is not something that is saved in this century data, whereas hopefully the.

C

B

But I I guess I, don't actually know for sure what's there, but if the merge request is something that we need to sort of cross-reference with in gitlab, then it's going to be very hard or not worth trying to filter by it here. If we're not saving it with the record I.

E

Have a suggestion: what if we broke down this issue into four different pieces like the search seems like the easiest? One could be a good first iteration and then I think we could possibly do two of those in the first milestone and we have time we can do three. That way we can continually ship.

A

That sounds great I think that let's go through the next issue, which is pretty well defined and then, in the latter part of this meeting, we can create those and get them put in twelve five and then take a more realistic look at capacity.

A

A

And the remainder one just kidding you.

A

Is manually opening an issue from a century or so that's exactly what it sounds like someone's doing manual triaging. They identify something, that's critical and they want to create something to track work with to help facilitate triage and speed up response.

D

A

D

About adding a button to the row to add a new issue and in terms of the format like what would be pulled into the new issue, we're still trying to figure that piece out but we'd like to follow the format that's already existing. When you create an issue from sentry in get lab, we have. We have examples of that. So we'd like to follow that format, but we're still trying to figure out exactly what that would mean and I think also out of scope for this.

D

One would be like attaching that error to an existing issue, though that would probably be something that would be nice to add in the future.

A

That's a great idea: let's get an issue for that.

A

C

Thinking about the workflow of having a button on every row, I think the button would then take you to a new page, so you can only ever click one of these buttons but we're having them on every row. Would it make more sense to just have the new issue on the detail page rather than list page all we want to be able to quickly add it just for an issue from the list.

D

That's a good question: I, don't know if you wouldn't have enough information from this list page and the idea was that you could, if he knew something was important. Just by seeing the title you could you know without having to go to the detail. Page just create an issue, but perhaps there is never enough information here on this page then you would always need to go to the detail page. First, that's good. A good, open question.

E

Yes, our existing team members to see whether that's useful or not as a first pass.

A

That's a great suggestion yeah, so you can absolutely put it in front of you. Guys. I mean one of the nice things about this product category. Is the entire engineering team will get to dog, treat it so we have a plethora of venice' feedback.

A

B

Perhaps since we don't know they and maybe to Simon's point what, if the first iteration was just adding the new issue button onto the detail, page.

A

I'm very happy with that Amelia. What do you think yeah.

D

That's fine until we find out more information and I think we're still trying to figure out how all of those buttons will work, because there's many buttons and where they should all live, how they should interact, there's still time another question that would probably be a safe way forward.

A

Awesome so I want to recap: this is a great great review from the top just to go through things that we want on the breakdown or punt on and we're all gonna pick an issue, and do it right now century to cluster well-defined will require some investigation, but it in and of itself can remain a single issue.

A

Great detailed view of first century areas and get lab needs to be broken out.

A

I mean at a very first past: we have improving the layout and look of the existing list, and then we have the error detail page in and of itself, so those could be there.

A

Then we have the addition of the issue button here. So creating an issue from an error just from the air detail page will be an additional issue.

A

Do we want to create an issue for the creation of issues from the list of you or shall we wait and see if that's a pattern that we want.

B

A

B

Say that we wait right.

A

Okay, great also.

B

On the detail, page other issues will be well. I think this is what we talked about. Adding the error details would be a follow up issue and then adding the stack trace would be a follow up issue, but first we would spike that out. Yep.

A

Total and so we have embed such a stack trace and an issue over here already created that one tool.

A

And then filtering the error list by release merge across and commit. We talked about splitting into you.

A

One per essentially ping that Amelia created so adding search or filter results.

A

And then adding to that release and or the larger breasts and/or the commit, and so from that we have for comment, did I capture all those ideas. So great.

A

All right, let's I, will start breaking up issues for detailed view for century areas and get lab.

A

Do a couple. People want to take filter error by this one. We've got four issues to create there.

A

B

Take that great.

C

B

Should one of those issues supplant this issue and then there would be three more additional issues or is there we have some sort of parent-child this.

A

Sounds like a really good scenario for an epic. So if you want to start creating those issues, I'll promote this guy to an epic and we'll link them. Okay,.

B

A

Then manually open an issue from a century. Error will become an epoch as well and we'll have the era details page and then the list one.

A

So what I just said Simon do you want to create those two without work for you.

E

A

Sweet and I guess we don't need to sit and I'll create those together.

A

Why don't we create those, and then people, just slack them to me and I will organize them after the column? Awesome, okay and the last one. This is not product feature related. This is something that I tried to do myself and I need some help with. So we've got this health demo project that is in a reasonable state.

A

I used a project template for it. I use the spring app template, which is written in Java, which severely limited my ability to get help because a lot of people don't know Java. So I would like us to stand up a demo project for the operations section that we can use for testing existing features, features surance, screenshots, demoing, visualization, new ideas and so on and so forth.

A

We'll need a light, medium weight application that we can deploy and then integrate century with we'll get all other features integrated, but we'll start with sentry asset Celyn, we're working on that right now, I suggested a name for this. Let me fix the formatting here.

A

Which is tanuki incorporated and then our URL will be tanuki IMC, and then anybody can join it without having to ask.

C

E

Tracking and monitoring stuff will be public, it might be scoped to permissions. Okay,.

A

Maybe this is wrong. This is the wrong word to use I meant so that we don't have to add every member of the else team.

E

Yeah we'd have probably to put on the get lab or group, and then that would work. Okay,.

A

Perfect, so this is another one that I want to do and 12.4. So once we get all of those issues broken out and I've organized I'm, sorry until about 5:00 and I've organized them into epics, we will extract the MVC's for each one of these issues.

A

Anybody have any questions.

A

Fantastic, we'll do another final review. Next Thursday I think okay, awesome guys take care. Are we good evening and a good day for Sean and Simon.