GitLab Engineering Productivity Team, 12 May 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Quality Engineering Productivity Weekly 2020-05-12

Description

Weekly Engineering Productivity meeting from 2020-05-2012

A

B

C

Hey sorry, everyone but I'm a little I'm late.

C

Serious theory also.

C

Sounds great yeah I was I was trying to get food ready for for Mike. My kids did you already get started at all? Well, no, all right! Let's start with everyone's favorite snack food I! Don't just keep it simple. Today, mine is chips and salsa, I think I've eaten chips and salsa as long as I can remember, and very particular in my sauce, that has to be like a restaurant style and here in the US. That's like very smooth, so everything's like blended up no chunks.

C

No, nothing I do like the fruit salsa from time two times like I made a pineapple strawberry, jalapeno salsa on Saturday. That was good, but um chips and salsas at the top. For me Remy. What about you.

D

um Yeah I think for me um chips. Also.

D

We really love that and with my girlfriend I think every week we we have what we call an Appy ho, which is like you know before the before the meal, so like appetizers I would say and by the way I think in France. We learn that chips is crisps and I. Think in the in the UK. Actually that I call creeps, Chris I mean yeah anyway, I think it's always confusing for us, because friends, we say chips as well. So that's but I'm relieved that in the u.s.

D

you also tend to say ships, ships, yeah I, don't know it: snacks, yeah chips and without that I think that's yeah, something that we try to stop eating and we manage for two weeks and then yeah. It's too hard, yeah, not really like yeah, not really super original yeah.

D

B

Mines, it's not something I've always eaten I, actually discovered it, maybe about four or five years ago. It's these. Like polish wafers. There are just really kind of small light wavers, so they're, not too sweet, but there they would be considered a sweet snack rather than a savory, snack um yeah. So in the UK we call chips, crisps and we call like fries chips. So.

B

Yes, so but Australians call chips, so crisps, chips and then, like potato fries, hot chips, you have chips and hot chips, but yeah. Yes, crisps, crisps are more like potato based, but then you have like corn based snacks, like tortilla chips. You know which you dip in souther and stuff. So we have I, think Doritos they're, pretty good, like cheesy cheesy kind of flavored.

B

What kind of salsa salsa do you have then Kyle? Is it like? A tomato is Elsa.

C

Yeah yeah, like a tomato-based salsa to me that usually tomatoes, red onion, think like a pico de gallo, oh yeah, so like that and just all those ingredients, just I like pico on its own, but I like everything, just smooth so puree, all that and it's spicy or not spicy I, like spicy. Actually the salsa there's like piece is like a brand in the u.s..

C

It's like pretty prevalent in all the grocery stores, I like their hot salsa, and it had been out since, like coronavirus started and all the gross I like I, think I went to like three different grocery stores and over the last month and a half and it's been out everywhere. My in-laws found three jars of it and brought them this weekend to me. So I finally have my favorite salsa back. It's been awesome care package.

B

Yeah yeah I find that there's been um snacks are kind of one of those things that you go to in time. Well and I found like here I've been stocking up on those snacks that I like during this period, because we're not going shopping as much, but it's also something that I'd like to have, and it's kind of like a comfort thing for me. I guess so I've been trying to get more C's more of these wafers recently, probably more so than before. This the virus started.

B

I'll bet. What about you.

E

I would say a banana okay.

E

E

That way, we don't look, we don't always be happy, then I realize that I can go through them very quickly and sometimes I want to keep bananas, so they not Quentin I can use them a lot of things like, but I don't bring. You know muffins or something but oh I, know going through heaven and Judas.

E

A

Really nice also, but not follow.

E

The light I want to change the light, but I can't get people to come to change.

A

So I think I also like chips and I. Have this it's not made by potato, but it says P crackers and it's use is very different. I think keep that I, don't know the name in English but different friends. It looks like this and basically this kind of package I also pretty much like.

A

So it's only I just realized. Everyone make sure this I.

B

Saw the video of the 7-eleven the chip selection in 7-eleven in in Taiwan, and it looks amazing like the amount of different chips you can buy. Yeah.

A

There's really a lot of different ice. Okay,.

E

A

You supermarket there also a lot of imported ones. Fish I, don't really find an orphan because they are mostly expensive and actually not very I. Think not many Taiwanese would like those flavors too. So it's mostly more local flavors, it's more much more popular okay and make may be welcome.

C

Welcome, what's.

F

My name there in the beginning, because I thought it was gonna, be a surprise. But then I saw my name there. I just.

C

Added there, when I saw you doing someone's.

F

Keyboard, it's really fast, so kudos that was it snacks I think I can just show you. This is a there's some snacks. On the background, this is like from from kovat 19, my my wife.

F

It's a big fan of Korean snacks, so uh this is, this is very famous in South. Korea is called choco pie and it's it's kind of like money in the in the military camps in South Korea. It's very valuable for some reason like you can get like ask for favors, and you pay your fellow comrades with with this and they're they're, really happy and yeah. We've got a bunch of other things that I did not buy this so for transparency and this Remy.

F

Don't don't don't be angry at me because it's French I'm sure it's not. It might not be kosher French. What is yeah? What.

D

Does it mean that doesn't mean anything.

D

F

What I thought.

D

Cook I, don't know this name best could be like from a city but I, don't know this city, yeah, yeah I. Think it's just something that sounds like French.

F

Yeah you want to talk about snacks hit me up anytime, like the wife, the daughter is pretty.

C

That's a whole thanks for joining, hopefully uh everything's going. Okay, I know it's very early on the.

F

Deezee slides are done so I can have time to hang out with all you folks. It's why I jumped in and braids.

C

D

C

Well, we'll move into the announcements: I want to call attention and my son is running around so I apologize for the noise, but the next round of 360 feedback, but first for some of us is starting May 18th, there's a nice page in the handbook with a lot of details, including the timeline and a video by Rose that kind of describes the process. I would encourage everyone, watches there's an AMA tomorrow as well I linked to the docs, so that you have a question, feel free to add it.

C

If you're not able to attend, I can vocalize it for you, but I'm really excited to see how this process goes.

C

Awesome next announcement is a group. Conversation is Wednesday, I linked the slides here. Please review them, make any changes that you bike. If you have questions you can you can just we can discuss them in the slack thread, but I appreciate everyone's collaboration on these and to make them make them as good as they can be.

F

Since I'm here can I, just let you a team know so I added one more page for for each team. Actually, that captures just a team. Okay are so that that's slide 13. So we have the Oakley arts at the top at the department level and we track and that each team is have the other on one page, landing page for the for the okay ours, so that that should be clear. Now know where things land.

C

Yeah, thank you for doing that. That looks good yeah.

F

And there were some discussions on on slack earlier on I'm duplicating chaos. It's essentially a pass through we are, but we can assign one issue to multiple epochs. So it's there's this excerpt paperwork that um that I have to deal with, because we want to track it from what the chief of staff it's asking us to do this time, so um yeah I I, am planning to give some feedback to the p.m. because this is a this is gonna, be our process going forward.

F

One thing we could do is just just resolve or mock mic ers duplicate, but I I was hoping that it doesn't close the issue. Obviously it closes the issue so that might cause confusion as well. So right now, I just have to see it open and just link liberally down down down down down to the team care. Yeah.

C

And and Remy Albert you, you both have provided good feedback throughout this whole process. I I know that there's room for improvement to make it clearer the thread and slack Remy I know you just added a bunch haven't been able to read it yet, but let's keep the ideas coming on how we can improve this and make it clearer, because there's definitely room for improvement.

C

D

Yeah one one small thing: I was thinking like without changing the whole thing, but just like I saw that the CEO chaos, our name like actually see o KO in the title and that's clearer, and that makes that clear. That is a clear for the CEO, specifically and I. Think in our case, maybe we should like add quality director care care, because I think that what confused me is that I didn't know.

D

If that was something I needed to look at or not like, and at least if it's written quality director care, I wouldn't bother looking at it. Maybe.

F

That's a great suggestion, like every everybody across the company has like K R, K R, so it's kind of confusing right now, yeah, that's that's something we could do and just check it that we're Thanks and to echo what caused that it's still very much an experiment. I personally, like the older from older format, however, we were trying out what the what they the overall process would be going forward and then Sid wants us to use epics I. Think that's also the key and like linking.

D

F

D

I think I think that's that's that works well. Yeah, I think.

D

The only thing that I find strange out is duplicated chaos, like that true tales and I still don't understand why you cannot just be assigned to the bottom letter, K I would say and an iPhone that in out n book page we actually we actually say that we shouldn't duplicate issues and just use the top like the bottom one issue: oh yeah, I yeah, that's still not clear for me, but I mean personally I don't have to care, because I just should just look at the team. Okay else, yeah.

C

So I would say we also care about the efficiency Department kr, because our team is responsible for providing those results. So I do want to try to be clear. There I know it's a director level and that's where I think a lot of this confusion started is but I'll say. That's where it's confusing for me, I should be clear in my language here that there's a department level K are the our team is responsible for that's not at our team of all were partially in our team level through duplication, so.

C

We'll we'll look at how we can clarify that I think I as well. Like I said you have some good suggestions, I've seen the slack thread and haven't necessarily read through those I like the pass through Kay. Ours are like one thing on that that maybe met can provide some more detail into it. That's that's something that we're trying out in the engineering organization from Eric. So that's an ass from Eric that we're trying out throughout the whole organization.

C

But yeah I appreciate the feedback we we can always try to make this clear. I just don't have answers right now. So, like let's work towards a cool anything else, I did move that down to the discussion, since it turned into a discussion trying to capture some of the some of the items here.

C

Great ok, this is kind of an announcement discussion, I put it down here in case there was questions I'm trying to be more personable in setting due dates for follow-ups, so I may set a due date on an issue where it's, let's come up with a plan by X date, using milestones to indicate when we have a planned release for something: that's not necessarily in progress or above anyone's cut line and then, most importantly, spreading non urgent work within a work stream that maybe one team member has been focused on to other members of the team to help promote knowledge, sharing and just kind of variety of work.

C

My goals with all this is to try to be clear in communication. You know like have a drive accountability with the milestone and create that sense of urgency and increase the opportunity for everyone to have knowledge, transfer and learn new things. I'm, not trying to add more stress through these. So if due dates are settings like creating this strong sense of urgency, that's adding stress, please, let me know, or any of these actions are confusing or adding stress. Please, please pass that feedback along any questions on that item.

D

Do you have any example of such issues? Yeah.

C

I can add them the security, so I'll say that the two issues where there are some examples. Here's the security MRI said a do data on to say: let's follow up on this plan by this date or, let's see if we can have a better understanding of the the scope by this date and setting a milestone just to kind of set. A target. Chinchin was great to provide feedback that where he said this isn't necessarily achievable so on setting that that was I.

C

Think the best example of doing these and then adjusting based on team feedback. I'll link to that issue in here.

D

C

The next item- I'll just kind of get this started. I- was curious, with the team's thoughts were on trying out merge strain for, like short, durations, very similar to what was done on the handbook repo. So we can have a coordinated, maybe an hour two hours and then look at ramping up to see what the impacts are.

C

Remy. You wanna, you want to leave this off. I see you editor yeah.

D

So yeah she knew I, replied and yeah. Basically, we have the problem of folks that don't run with like that runs in the in their own namespace I would say, like trough, talk pipelines, so so, basically for community contributions that wouldn't work, um I'm, not sure if that Isabel's mesh, like I, think we can still merge, but that bypass and make the train derailed. If I can say that, and so that would like remove all the current node requests from the Train.

D

So if we, if we still want to experiment I, think we should like tell maintenance to not merge any community conclusions during the experimentation, so that just something to keep in mind and to also to confirm with Shinya, probably.

E

You see the drill, does that mean that it doesn't run? Because is that? Because we can run some pipelines on the phone or is it because they've emerged? My stream has to be restarted.

D

Yeah I think I think yeah. The problem is that there's a security issue with running.

D

External contributors like running external computers, codes in the context of the canonical project, if the logic quest like exposes any environment variables, I think that's that's the issue.

A

So it makes sense if we always create a branch in the canonical which has for any community contribution and merge from there, because it's it's it's an extra work, but probably not too many extra or because we just need to do it when it's ready to be merged and that avoids all kinds of fork. Issues.

D

C

Yeah I'm not I'm, not sure, if there's an issue for that I know that has come up in the past, because the same challenge that we talked about, where jobs that used privilege like tokens or anything like that aren't run in forts I thought there was an issue for that.

C

C

The thing I guess the context I didn't provide as I was hunting for the issues. The things that I'd be looking to understand better by experimenting is like any like how how often does the merge trinket reset from like flaky pipelines, so how reliable is it what's the performance of our master pipelines? How would how would that change from the car state and then Remmy? One of the things you brought up yesterday when we chat about this was what does the pipeline graph look like for the subsequent Mrs that are in the pipeline?

C

So if you have a docks pipeline, that's first in line and a pipeline get started for the second one. That's like a code change. Does that also include the docks jobs and the code change? Does it just do the code changes? Those are that there's there's some questions on. How do all the rules like configuration rules in our pipeline kind of overlay with merge trains where there could be really valuable feedback for the verify team, just even through the experimentation by the loan? It's master stability.

C

So it doesn't sound like there's any reason other than the fork pipelines which looks like it's scheduled and lurching one. So we can work around that by encouraging maintain errors to not merge prime ours from forts during that time, but it doesn't sound like there's a reason not to try or not to consider trying.

C

Okay, alright, so I will go back, I, see I'm, you real quickly, I'll go through and do a plan to, or just like a proposal on that issue and get some more feedback from the team async before we look to do it, it's not anything that we'd look to do like in this release, maybe in the middle to the end of thirteen one. So it's not something that I want to set high urgency on just wanted to discuss, go ahead. Mac is.

F

It is it because, so if we can't use motions all in one, I assume the merge results pipeline and merge trains can't coexist in the same implementation. Can it meaning the ones from the one from our team members go through the merge train but then anything from Forks? Can we use convinced to use just merge, super spy apply and they they just go in another pathway, not not cool, and we had a train.

D

I'm not sure I'd I think any time you merge directly to master without the mass train it it resets.

F

That's the drill to train, that's what we would have.

F

Yeah the feature is working as intended: I see, okay, yeah I think yeah go ahead. No.

D

Our just thing with the issue that that, or it's mentioned, that that's scheduled for thirteen one I, think that should solve the problem, because that would allow us maintenance and developers in general of the canonical project to start a pipeline into Connecticut projects after reviewing the code. So after checking that the code is fine and not exposing variables, so I think that would solve the program and I.

D

The only thing that we do you would need to check is that starting a pipeline manually is fine, but it would also work with the measuring which starts the bike runs automatically.

E

D

I think I think that would solve it since that, basically, the problem that this needs to solve so.

F

Okay, yeah makes sense, III echo what you said to Kyle lower priority at the ie I. Think even 13.2 is fine, since we have a lot of momentum early on costs. Just keep on that momentum. That's gonna help us a lot going forward. Sure.

C

Okay, cool. Thank you for.

F

The discussion, however,.

C

You want to take the next item. Yeah.

E

So I've been looking into the front end fixtures job recently. One thing I found is that this pictures are mostly JSON and HTML fixtures. They're static did so they are taking a long time because it's running yes, our state test, which creates the data within two hours by yourself, like that using factories and stuff, and so whatever also using the background jobs, which then get repeatedly called in accumulates time.

E

So I feel like it's. In most cases, it doesn't need to be run repeatedly. So I was thinking whether we can use therapies, fixtures or something different or any other solution that we can to recreate this to get to the same. In result of having the data ready for a controller to render JSON curious and water I guess everyone else.

D

Yeah I don't have a good idea right now, but yeah. The only thing is that I always thought that yeah like using a spec to generate fixtures is on one hand it's knit and on the other and it's a hack so and for very complex features like the psycho analytics. It really requires a deep knowledge of how all this data is like what the structure of this data needs to be before you can improve things. Yeah, yeah I, don't have a good solution.

A

So what are we expecting from running lives? Are they generating a lot of JSON files.

E

Yeah, the most the old JSON and HTML files, so it calls a controller and then renders the result. The JSON saves it into a fall and I get the JSON files get stored, but if it gets used in a mix job by the JavaScript test which of these forms and fixtures, oh.

A

So there's some older in dependency.

E

Not in no I don't think there's any order in order dependency across the different votes. Then.

A

Can we maybe generate the Jason's and commit the fires, so we just believe from time to time.

E

Yeah I think that's one solution, but I remember seeing a comment somewhere on the issue before where it was not recommended to commit Jason.

E

D

That I think I think that what we need wanted to.

E

D

He left- and that's actually true to do that like commit the fixtures like generate them dynamically, that, like we do but then community, so that we don't generate them dynamically in the pipelines, but every time we need to update them. We can do that locally and then do an odd request.

D

I think that that was the plan and there were a few issues to get to that point, because the pictures need to like, like the kilchers, need to be like it impotent when you reach animate them like the dates, shouldn't change and stuff like that, but yeah I think that was thanks for reminding us, because I think that was the plan. So we should probably focus on that and see what is missing because.

A

F

Need it a great.

A

D

C

Too, do we know where the issue that winnie worked on wasps or is that definitely uh I can.

E

Find it okay, the.

C

Other thing I was gonna, throw out is, if you need like, since these are cycling analytics. If we just need help from that team, we could we can ask them, it might be low priority, but it doesn't hurt to ask and say hey: do you have any ideas, whether it's I'm not sure who the front end people aligned with with that group part, but we can ask there yeah I.

B

Think I found an epoch which is related.

C

So it sounds like review the plan when he had mentioned half cycle analytics. You need some help, thanks for sharing the epic mark and we'll move on.

C

Anything else Albert or the awesome great discussion on that as well. Thank you for bringing that up. Chinchin, you wanna! You want to talk through the last one. They camel to this design, define assignment physically yeah.

A

So last time, I move all the group definitions from the policy files to Ruby file and I try to move them into llamo, because we realize that there's a lot of mismatching with the current stage definitions, so I literally I want to somehow use a script to generate the fire and we can adding any more assignments to to generate fires if needed. So we don't have to memory, go through it all the time and and in order to do this, this is the first step to make it more like a stage demo file.

A

So we use the same key and use the same format when when when what weekend? So that's it, I already saw a lot a few comments. So thanks for the feedback.

C

Yeah great great refactoring here I think this will help with the specific requests on this Idol Mr triage report. I would lean towards having I'll say like a script or some sort of patching system, I think just using the stages or using the stages diamo as a single source of truth might be a little disruptive. Even though they're weekly reports is but something that we can work through on the issue.

C

B

Yeah I think it's a really good idea and it needed cleaning up for a long time. So, thanks for that.

C

All right well, I, have a co-pilot in my lap. Does anyone happy anything else for the rest of the to finish the meeting.

C

Okay and with that I think we will let everyone go I hope everyone has a good rest of your day and we will catch catch up the rest of the week async thanks for joining them.