GitLab Dev:Manage/Plan/Ecosystem UX Design Reviews, 9 Aug 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Dev-Manage/Plan/Ecosystem UX Design Review - 2021-08-09

Description

Product Designer for - Manage:Access; Daniel Mora, speaks on two topics Access is working on, Service Accounts and Workspaces.

A

A

Add this to the.

A

A

So I don't know if you have access to the agenda or not so I'll put it in the chat.

A

So I'm working on service accounts or bot accounts and I'll share my screen also real, quick.

A

Let's see it's this one.

A

No, that's not right.

A

Where is the screen.

A

There we go uh so this is something that's been going on um for quite a while. Actually we want to do access tokens or specifically, bots. um There's been not this issue.

A

Perhaps it's the previous one uh yeah this is it there's tons of documentation and comments, and back and forth um become a bit of an emotional issue for some folks, just in terms of the the lack of implementation of that lack of execution in the perfect iteration of this versus nbc or small iterations, okay, and so I was using this kind of as a reference point to do. Some research or study around other organizations that do bots or service accounts.

B

Oh, my god, I was working on service accounts at my last job when I left.

A

Perfect uh so you'll be able to have some feedback um so generally, what I was looking at was my first iteration of um what would be the idea.

A

Could we have an idea if we make a new bot button versus a new user, uh and that looks like actually, I'm not sure if you've got installed the gdk just yet, um if not no worries um you can get to it at some point in time in the future.

A

But let's see so the idea is, you would click on the new bot uh and it would have kind of just generic information that currently exists within the new user, but this is kind of, like I said, a first iteration and I asked what sort of features or points are relevant or not necessary.

B

So is it, is it always a it's? Is it always a bot like what does bot mean to you, because.

A

My familiarity.

B

Service accounts is like a developer account or an integration.

A

Exactly so, what we're looking for? Isn't a few things let's see here.

A

So access tokens is our current iteration. Okay.

A

And we have access tokens, we have excuse me personal access, tokens, okay,.

B

A

Whenever this loads, we have project access tokens, we have, um let's see.

A

A

A

Deploy tokens and now there's a an issue to create group tokens and instance token, so, basically, access tokens at every level of our uh environment, I'm not sure what okay and so the idea is we're going to start with just personal access tokens for the first iteration and so from here. You just create the access token, as you normally would in our environment.

A

Once you generate the access token, it gives you the key or the token itself, and then you integrate that into whatever your code you need, and so the proposal that I had as well can we make this a little bit easier? Is there some way we can make this not as complicated or more flexible, because the limitation we have right now is just strictly at the um the object level. So in case you know the project access token or the future, the group access token or the personal access token or deploy keys.

A

Is there a way to kind of make this logic more streamlined, and that's where I was trying to go back and research or look at other organizations how they implement them?

A

So it came down to this as the mvc, where um I think this is it that's the closest we are missing.

A

Yeah access level tokens should be configurable instead of being maintainers service accounts. Tokens should be somewhat independent. Allow the creation of multiple service accounts, allow creation in multiple access tokens and varied.

B

A

So, and that's where I was getting with this kind of first iteration like ugly wireframe, um but then I was thinking to myself. Well, I have this idea where this is the new user creation.

A

Okay, if look here I'll go back a step, the initial or one of the initial asks was that there's no way to see what's a bot and versus what's a user in this screen, so we were going to say why don't we just drop another tab in here to separate? What's a bot and what's a user.

B

A

And so I was thinking well, let's drop in a second button here that says, create new bot versus new user, and then I thought well, maybe, instead of just doing that, we just go to the new user section so we're not having to create a whole new page from scratch.

B

Our what percentage of our customers have at least one bot.

A

uh I do not know yet so that's something that I would have to further investigate, uh but from here I'd want to add like another feature that says what kind of user you want to create, and so the idea would be either you create like a human user or a bot user, and if you select a bot user, it would get rid of things that aren't necessary. So these containers just disappear and it would only have things would be relevant. So in this case you know access level uh things like that.

A

So that was my first or excuse me. My follow-up to this topic that we have here.

A

There's quite a bit of discussion here in this thread: uh let's see so yeah, it was this right here and I was trying to create a iteration off of that and I think we might have like the first thing to prototype or test with based off of that logic.

A

I'm just waiting for feedback from imra there's a little prototype link itself, um but yeah. So the idea being that I think that would be the first mvc where uh oops, so we create a new user. Here's.

B

A

Type and then it would define it and then you would just save it, um but again like I said this is just kind of a first iteration. I wanted to get feedback on this and see if this made sense so.

B

Are you using bot and service account interchangeably.

A

Yeah! Sorry! Sorry about that! No no.

B

That's okay! As the newbie I like all these things, you just kind of know. I don't know.

A

Any of them, I'm also learning right now, actually on the fly, because I'm not usually, uh I haven't really messed or investigated with service accounts and bots. Just yet, okay.

B

A

My only experience with them is like what we have here in our environment, um so we have some like uh gitlab support, bot, gitlab alert, bot, there's some other ones that are that already exist from the gitlab organization that we offered to protect our environment, so somebody uploads malicious code. We have a thing that would do like um check for that and say: oh no! This is malicious code. We're gonna flag! This. You can't integrate that code into our into our environment.

B

A

So we have defined bots uh end users. Organizations would probably like to generate their own sort of bots, and the mvc would be to create personal access tokens for the admin or end user. Whoever is generating the personal access token go into this little prototype, create the bot user.

A

The surface account define its access role and then save it and then go into its code base or the the person's project and then integrate the token that they have generated into their project and that would generate the quote-unquote service counter bot, and this is the first iteration that we have, that we want to try and prepare.

B

So somewhere, along the way, the access token is, is it assigned to the service account or how does that work?.

A

That is exactly the question I had. I was asking: how does how do these two pieces communicate with one another? uh Where did my discussion topic? Go.

A

And uh imra he's uh our back end engineer, that's working on this with me: okay, he stated that that would be done in the developer's code or in their environment. um That would be done in the back end. So here I said, creating the service account would be just giving it an access level. I assume what comes after this is in terms of linking the two after the pnts are created, uh it's done in the devs code and not in the in our environment and so yeah.

A

That was the uh the way they would be done.

B

A

So again, like I said, this is all just our first iteration, with kind of a first investigation on to how this problem could be solved. uh Okay, and so my next step is to kind of maybe try and start building out some visuals and like have some feedback sessions with um some users just to.

B

A

Solution, validation, um rather than trying to develop this and going straight into code. Yeah.

B

A

This one is slated for q4 uh and they have pushed for q3 being workspaces. um I.

B

A

Because I I don't so I obviously I guess you know you're gonna be in workspace and not access anymore, because it actually just has been broken up a bit.

B

Well, I don't know that because what ori told me is that they're going to be breaking it into two, but I can pick which one I want to do.

A

B

I don't know for sure, that's just what she said. One time briefly,.

A

Gotcha, so I don't know either I'm trying to figure that out. Also um because, like I said I used to, we used to be access, and now it's gonna be workspaces and something else, and I actually think it's going to be a third thing.

A

My initial understanding was workspaces access, like validation, login, and then there was a third item which escapes me right now, but okay.

B

A

Else- um and so this was like earlier discussions, so just in terms of what you want to do or where you plan on wanting to put your expertise in this might not be relevant, but this is kind of what the task that I'm working on right now.

B

Okay, gotcha yeah this was this was really interesting. It's a very familiar problem to me so um yeah and I wrote down the issue number and I'll go read. I want to read the problem statement and everything again.

B

A

Yeah, so this one and workspaces, like I said, are the two that I'm focusing on. um I think I tagged you in the workspace stuff also that one, I would argue, is probably more further along in iteration because it already exists in the gdk, which is this environment here.

A

So basically I have a localized instance of gitlab, okay and if you wanna, we can talk about that now, since I don't have any alexis took off, she said um so yeah, it's a localized installation and we wanna bring features specifically like this called admin into the self-managed.

B

A

Excuse me, the sas environment, which would be this.

A

So, let's see oh lord, let's just use a project.

A

So from here there's features and settings that don't exist within or excuse me yeah. They don't exist versus what we have in a self-managed environment. So here I have this thing called admin yep, where I can go and have administrative features across the entire environment.

A

However, a lot of these things are hardware, environment or install base environment, questions or features which don't necessarily have any relevance here in the self management or excuse me, the sas environment, because obviously we're not a sas customer is not going to be trying to play with our hardware right. That would just be a security problem.

B

A

However, we want to offer them things like this, for example user management which doesn't exist in this environment. I don't have a way to manage the users for my organization. I have what's here membership for a project, but that doesn't necessarily mean that they are part of the organization.

A

So there could be team members who come in through, like a saml, uh a sample group sync uh that don't necessarily exist in this project because they're in the different team or whatever, but they exist in my organization, and so because I don't have that way to look at everything from the top level which this admin area does we want to bring that feature over to this uh self-managed environment so that I can look at all of my users and not just the ones in a specific group.

A

So from this level of visibility, it's hard to kind of visualize what it looks like or the topology, and so this is. This is kind of what we're trying to define and so the best way that I could kind of visualize it or explain it. Is that if you look here, you have things like milestones, snippets admin, these don't exist in a hierarchical logic, structure versus groups- let's see here so arguably this screen would be.

A

The logical container would uh not actually- uh and these will be the sub items for that logical container and that's the parallel would be here, for example,.

A

So this would be gitlab and you have everybody underneath it all. These sub groups, sub projects.

A

However, this exists at a layer above so arguably the layer you're seeing here. Gitlab.Com would only be this, for example, okay and you wouldn't be able to see anything external to it, which is what these are and that's the workspace environment that we're trying to find in the self or the sas model. So.

B

Who creates these external groups? How do they get created.

A

So the way that would work is, if you look at the sas environment, think of it just like this, so I would just create more new projects or new groups from within here, so anyone could theoretically do it. The problem is in this environment. You wouldn't necessarily have access to them as a whole or as a totality, so say. For example, here you have the whole totality of the environment and we want to define or add a feature like. I want to see all the milestones for all these projects, all these groups in totality.

A

So I can measure my organization measures across versus just one container object and that's where this workspace environment is the container that would have to have all of that. Those feature sets and again.

B

So it's going to be called workspace in sas and will the name also change in self-managed.

A

Yeah, it's hard to say we don't have. We haven't discussed that right so and even still the workspace was kind of just like a band-aid topic. um It was kind of like brought up for discussion amongst a few people, but it wasn't measured against march market uh testing or anything like that.

B

There was discussion.

A

About using certain things, but there was like no it's too close to competitors and arguably there's possibility for that to be further explored, but we're just calling it workspace for the time being just to kind of move forward. um Okay, I I don't know, imagine there'll be much pushback on that. Anyways, though.

A

So hopefully, that gives a little bit of insight as to what we're doing with um the workspace stuff and again going back to what I was saying about the mvc, for that was that the one of the initial proposals was doing specifically user administration, because that doesn't exist. The time being, there's also uh an mvc proposal that people want that they want to be able to see groups and um issues epics.

A

um I was arguing against that because I haven't started or investigated anything around that and it's too bad alexis dropped off because she's more attached to that part of the organization than I am, but the idea being that they would also exist just like, uh let's see epic, so here I'm at the top level. I have epics and I have lists- and I have my boards, but I only have them vertically within this organization. I don't have that insight laterally across the totality of my organization structure,.

B

A

And that's one of the things the other embassy that they were trying to propose for that.

B

A

I don't know looking.

B

At an issue, is there an easy way to find what was decided on as the mvc or is it just like hidden in a comment somewhere.

A

uh Let me go back and get that for.

A

A

A

B

A

I believe I tagged you in this one, so you have it for fyi okay, uh so you can, if you want to poke at that, uh but this has a really in-depth discussion in regards to like the technical limitations, architecture mapping, um you know the work efforts involved, um let's go back to the top um the breakdown of how they want to try and plan this or investigate.

A

uh This is the discussion regarding the proposal.

B

A

This was my iteration one proposal where I was saying we're just in bringing in the users and the work that's already in progress for that uh iteration two or 1b I should say, is groups.

A

I argue that that was low value or I should say.

A

Less feature usability like it's just the group insights, like okay cool, I can see whatever's at the top level, but what do I do with that information? uh And there was another.

A

One yeah so aurit suggested that we bring plan, features uh issues and boards epics as an nvc and that's what I was saying. um I didn't necessarily have investigation into that and that's where I commented holly and alexis uh if they could bring any sort of um discussion into that insight. Okay, um uh would they want to talk about that so holly gave some insight.

A

Gabe is the pm for that group um regarding issues and boards and epics uh so yeah here. This is just a higher zoomed out view the problem, one of the issues again going at the scope. Whenever you go higher, you have much more content you're dealing with, and so we have a performance issue with population of that content. So for.

A

B

A

Yeah so epic, so we have 4100 epics and then, if you imagine, you want to populate the that laterally across your organization. How many more is that going to explode to and then, if you think about that vertically, so the next thing is epics would be issues so imagine how many 10 000 issues you have per every one of those. You know theoretical epics.

B

A

There's a huge mapping problem and uh problem with getting that data to the front end without blowing something up. Okay, where was that discussion? Point yeah uh have performance issues, so these are all things that we need to investigate on the back end um so proposing that as an mvc, I think that's the concern is what's the issue that would be less problematic or require less effort from the front-end slash back-end.

A

You know any other part of the organization, whereas what I was thinking about the the 1a mvc with just bringing user admin features to the front, or this 1b, where we just have visibility across the groups might be less of a team effort and more so just that the front end can just kind of smash out an iteration on that since the art, the logic already exists in the self-managed environment.

B

A

Just uh we don't have an interface or a window for that.

B

A

And so that's kind of where we're at right. Now we don't have an official first iteration mvc. We have a couple of proposals, I'm moving forward with this thing that I was speaking of earlier with um user management.

A

This is kind of a separate issue that started with this thing called view and set attributes, but the mvc for that is to have visibility to your user management.

A

The first proposal to get rid of the um admin features of it, just strictly the visibility. So just I want to see the users that I have in my organization as the first iteration to test the logic, uh so no admin edit, no new user creation.

B

A

um So yeah that's kind of where we're at with workspaces.

B

A

And so yeah, like I said uh just in terms of like how you feel moving forward, what you want to spend your energy with what you feel your expertise lies in.

A

Don't be shy about voicing your your opinion, your concern or pushback, um and I would say you know you have you the expertise you need to solve all the problems versus myself, I'm relying on you to help me.

B

Well I'll do my best. I.

A

B

So much to learn but uh we'll get there right.

A

Yeah for sure, and I'm just kind of going back outside of work, but more so specifically around the task of the onboarding stuff, like yeah.

B

A

Didn't finish my onboarding stuff for like months after so don't worry, okay, I would say focus on like some of the basic stuff, but then mostly stuff around. You want to work on um for the team for what you want to proceed with for your projects. um I'm not sure what the pm onboarding stuff is, but ux stuff was a lot of like reading the handbook reading, like uh about the organization reading about the technology learning and how the the environment set up yeah. So things like that.

B

Yep yeah, the pm stuff seems similar, so well that's good to know that I don't need to chug through it. I can gradually pick up some. You know more access, specific stuff, rather than just.

A

Yeah definitely focus on stuff that you're going to be working on.

B

A

Just kind of like quickly skim through the rest of the stuff because, like I said, you necessarily won't, have direct contact with that apart from chatting with other pms about hey we're doing this feature. Do you all want to you know partner with it or is there anything we're gonna any toes we're gonna step on.

B

Okay, all right awesome. Well, thank you for the advice.

A

Yeah yeah we're at a time. um I'm sorry, I didn't have any space for you for questions. Do you have any.

B

Well, I felt like we were kind of going back and forth and I was asking questions.

A

So, okay, perfect for now.

B

I got them all answered and you seem like a great resource, so I will definitely be in touch with more.

A

Yeah yeah anytime, you want to like chat about stuff or have any questions just shoot me. A message um like we got our meeting scheduled next week or something I don't know, but yeah I'll definitely hang out and we can chat.

B

All right thanks daniel cool.

A

Good to see you take it easy bye.