GitLab Monitor:APM Team, 22 Jan 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Monitor:APM Weekly Meeting 2020-01-22

Description

Weekly meeting for the Monitor:APM group. Discussed 12.8 issues, 12.9 planning, logging, and some issues.

A

All right tell everyone: this is the weekly meeting for the monitor a PM group. Doh. Welcome back hope. You had a great experience well,.

B

It was quite an experience. Yeah I may say that you know if any of you think about participating in the CEO. Surely problem I would highly recommend doing so, and if you have questions around that, if you can to know how we're thinking about it know reach out to me.

C

Are you all well rested now wander.

B

B

Catching up with so many.

A

C

D

A

D

Sure yeah I missed you.

A

Alright, let's get started with the agenda, then so Dobby of the first handful of things yeah.

B

Thank you for doing that. There is a discussion about what are the things that we think should be in so I encourage. All of you contribute so you to go way to do it. Put your comments, a few things so that yeah, let's have a conversation, they're.

D

A

Hidden questioner yeah so, as part of that comment created a kind of a thread in the discussion asking all the people to chime in with issues so I created one this morning my morning too, for the backend a similar one. So if there's specific things that you want to work on either in front-end or back-end, just you know you can start putting them in those threads and that'll just be a good way for us to start tracking those.

A

Or feel free to just use that issue, however, you want, if there's, if you just basically, we just want to start collecting everyone's ideas, so if you want to put them in those threads great, if not somewhere else in that issue would be just as good yeah.

B

I think it makes sense to start like putting some sort of an end goal for this, because we want to make sure the issues are playing in time. So I think that the date for scheduled it is the front of the street.

B

But you know, let's give it like, maybe a week and we'll stop putting on the list of the issues.

A

It commits you had the next commentary: oh I,.

C

Mean it doesn't really have to be verbalized time.

E

C

I mean the plan is, is to have everyone contribute to part of the discussion, so everyone has more visibility to what we're doing yeah. The goal is that if we can all contribute a p.m. and if we can all collectively kind of share like what we think is important, we can also influence the direction everyone can. You know, participate in that and I'd like to see more of that, because I think that makes our team stronger. So that's kind of the purpose behind it.

B

Okay, good stories also so I have some had some conversation with some of the members here in the team.

B

Before the afterworld.

B

Matrix matrix, I think I, just it it's not it's an update, and so two things in mind. First of all, to be clear, this is not my page. This is our page. This is the team page, and this is our focus. So if you see something that you don't like, if you think that something is wrong, everybody does not certain stunts, go ahead and change it or merge request, and you know we can we can. We can collaborate on that. This is number one number two information about like the basic things on this racing.

B

This is no game. What this matrix, what we're working on, because.

B

Music monitoring, and without so it's important that all of us will be aligned technology.

B

The tools that we are using and so I encourage all of you to read that also math implement new people. The join make sure to have them that section that twist them have it the part of the reading materials from the hand workbook can be intimidating pages, but those are important pages, but I think all of us would need to know them and constantly update them, because this is something that we are working on. So we will change it. We will update what we achieve and what we completely to work on next make sense great.

B

B

So we, this is an example of the validation trick that I'm doing together with the UX team. You exist election designer.

B

We learn throughout the breakdown, so people.

B

61 people did also that same so I encourage you to look at survey results and the last thing really critical information along there. So take your time go to that. You know if you had. Some insights had some with my own, but if you have a system inside feel free to share them, you know either in the issue in slack.

B

Whatever you think is a perfect.

B

Okay, realize that you have a question here.

F

So I was looking at the survey results and the questions or what other questions said. How many applications do you use in production and my people are nine responses were at none, so I was confused if it meant that they do not have any applications at all, or does it mean they have it another day and but not in production? Yeah, probably.

B

Either they were tolling, oh, they may be they edit. They still don't have it in production units may not talking to people that are still working on creating their application, so I think it makes sense. You know out of 160 people walking.

B

Into things, so we go over the three nice and next one for me, so we have done exploring. Thank you all for working on that.

B

I want to understand what we think should be our next peyote I capture, some of their a discussion on snack on our channel. So I got you some of the answers here. A similar one idea is anything it's ready, but you know maybe we designs, we disagree. The DYP lastik search is not installed, so this is like one one option. The other one is to support a generation.

B

So is the fact that we, when we go into the mix.

B

It was like a age of date. You cannot.

B

B

Keep going and doesn't make sense that customers are hitting the war not like if they cannot go beyond that selected timeframe, not sure if they maybe, then we can do it, but it's not really intuitive. It's not really user friendly.

B

So I think this is probably one of the most important issues that I would like us to address for the next iteration. If we need more, hopefully to break it down like in the technical level, then we should go ahead and do that.

B

The next thing is the filter search. Okay. Now we are doing filter to the design issue.

B

How we can do that filter using the search panel similar to the way we are doing, searches filtering on planning mode and all the nits in the season understand its speed according to whatever stored I, know we're collecting blocks from all namespace.

B

Right now will not fit on my namespace, so we should do something around that, maybe in the filter central, because there will be a lot of other thing that would like to use this capability like defending. We know they are constantly asking me when we have this, you know Explorer ready and when can we like put some some more requirement because, as we collect in the data from all the instances, there are some data that is important for this team, not only for the application on antennas and so may be filtering by this thing.

B

G

It of just about the namespaces, so the issue is the current view of the logs is filtered by environments.

G

So an environment is the primary way you select your logs and so an environment is basically a couple cluster namespace, and so it's it's not really easy to add drop-down for namespaces in the current view, because you're already filtering by a namespace when you select an environment, and so this is something that has been discussed and maybe thinking about doing another view of logs in the cluster administration, where in that view we would have instead of environments, you could select any namespace, including apps, managed by git lab or any custom namespace that you have that doesn't map to an environment.

G

So it's a bit tricky to do the filtering by namespace on our current page. At the moment,.

B

Yeah so the question: is you say that when we.

B

Think by a single space.

G

Into an environment, an environment is a namespace in a cluster. So when you select an environment, you select a cluster in your selected namespace.

G

So so you can't really have both next to each other, because that doesn't really work. So we either need to change the way logs are currently presented and use namespaces in that view as well. But then it conflicts with other functionalities that use environments or we need to do another view, for example, in the cluster administration, where there you could, it would make sense to have a view where you could view any namespace, of course, you're already administrating the cluster.

G

B

Yeah yeah, that's an interesting one. Yes, Misun start an issue like a discussion issue. That means in freaking to somebody is either like hey the news.

B

Something anyone need to think about we're, not so big but thinking you should make sense sure. Let's support Thanks.

B

But you will feel the stage boy.

B

Okay, so yeah we put on the discussion the discussion in the in the 12.9 planning issue.

B

I'm so forum slack, you want to add anything else. I mean want well done. We mentioned the UI.

B

G

Yeah pagination, so for 12.8 we have a refactor that is scheduled because the code is currently in mess and some other people want to use it.

G

So it should make it nice for them, and then the next thing for me is this performance issues specifically when you navigate when you select new environments and new parts, every query that you do happens in background workers, and so that has like two three seconds of latency every time you change something in the UI, which is not really user, friendly, I, think, and so we should probably think about how to improve that.

G

There's, it's not an easy fix, but we should probably discuss it is.

B

There an issues role for both items.

G

Not for the performance issue for the refactoring, yes, there is, can you create the issue.

B

Cool Thank You.

C

D Clemente yeah I was just trying to see you know what the so one of the things from the engineering level the director level seems like they want to try to improve or dogfooding. So it's just trying to see you know in light of that, like maybe that can help position figure out which features with your prioritize but they're, just throwing out some botch related to that, since I think we use Cuban Cabana for us Ari's.

B

Yes, I actually would love to to see how we can vote for this feature because we're definitely in there, as you can see, there are so many places for us to grow and evolve well, it will be nice like and we know well, we know what we want to fix, but after that for me, users- and you know who is using Cabana, like which team yeah.

C

I think the infrastructure team uses it for debugging things, but I'm not sure specific people, yeah.

B

The thing is that I mean we are using well, the plainest exchange to a cluster I think that what those team are doing and I guess there are several uses for Cuban. They actually monitoring blogs.

B

External sources from did lab itself so for them, starting to use our Explorer would need to support an external elasticsearch to connect to that Nome Explorer, because I'm not sure that them, but then it maybe I'm wrong I'm, going to ask them, but I make sure that they are using the elasticsearch that is deployed within cluster.

B

So this is why it's different I can I can maybe try to engage conversation and turn and see if it makes sense for them to start putting this. There are monitoring any application. That is you.

G

For your friends, stuff, I I, linked in the in the document and pre-existing issue where we were discussing some recruiting solutions.

G

B

Okay, yeah thanks.

B

F

F

Yeah, this is an aberration that we have, that is related to long Explorer, which is already freshened I. Just thought, I'd mention it there. That is still on the next high priority list, because it is scheduled for 2008, and if it is not that important right now- and maybe we can push it down, for instance,.

B

So I think I mean what's what's the status with the Refresh itical.

F

Yeah I think it still needs some back and work and back and much liquid as well I guess, but it hasn't gone in so once that goes inge and on the front it.

B

But right now, what's the status today is it refreshing is not refreshing. This page.

B

Yeah, you don't just.

F

Reload, the entire page.

B

B

Because this is my take, so it's good that it's included eight.

G

Yeah, thanks for raising this I think I also forgotten about it. So, but we should probably do something for refreshing.

B

B

Okay, next so I like you, where several events.

B

That I'm elated to be a monitor dilemma is one example, but it's also cute Kron I'd like to see if we can maybe submit some talks to one of those events, I think it's I think this team probably has many issues that we can invent or that we can submit considering the way that we touch elasticsearch Prometheus and some experience with the air there. So this whole those whole set of tool.

B

Actually what the sponsor of those events are looking for, because they are they all open source to.

B

Project and comatose so you've never interesting story to tell you know. For instance, we just elastic monitoring, pretty sure that you know we have some anecdotes to tender. Music colleges all day, probably have some things to talk about, so I would highly recommend you to take like maybe like 30 minutes. Think about an idea. We can.

B

You can talk me into discussions when we can brainstorm on that together and I think it makes more sense that their game will come from the engineers that are working on the tools they not formed from the product team and but I would be more than happy to help you to support I. Think it's good first for from good lab for the team, but also for yourself to start submitting talks, and maybe, if you get selected, you can go and speak.

B

If you are afraid of speaking and you know, I can can be there to support you and if you have an idea, you don't want to do that. Talk to me and you know I, don't like doing that. But if you think that you don't want to go up on stage, maybe you can't act in with someone you need to do it together. So I think it's it's a huge benefit for the student. Has this ability to destruction? You actually think about those tight tight. So there is an issue that you can continue.

B

You can have your comments there and as I mention, if you think of something- and you want to reach out to me slack- we can properly call we can brainstorm together. I can help you do that thanks. Everyone talks in the past not to dis event, so I can understand what they are looking for. So would love to see it happen.

B

Fiona verbalize a question or supplement. You know first.

B

I meant you them yeah.

C

I, don't really need to be boys, okay,.

F

Oh sorry, I'm just going for money, a great point which is that the scheme that recorded and looks like you went found something.

B

Yeah, it is recorded.

B

But it's similar to like most of the other event. You know you submit talks. The probability for you get accepted is not that high, not because you do, because there are a lot of submission. I have to say that mini like like Bill beautiful talk not to talk about the company, and you know what great both will do it, but talk about the product. You know this is why I'm saying like like for polytheistic search was a popular topic.

B

So think about the means is that your head, once with this product some and then you can tell the whole story the K I'm walking on beat lab. This is what I do they need to do that I've done this. This was my best practice. This is experience that they have and so on, and so it's it's really. Nice actually go ahead and you can also go to some of the talks to better understand what is there?

B

What is the concept of those discussions?

B

Okay, I think I'm done.

B

H

Will take cover I have one point: I just want to rise. Their intention I had a Mazepa about how a PM team stores the dashboard templates so like the common metrics dot, yml Gabrielle express that that we are we that we store them in the config.

H

Slash Prometheus directory suggest that this file could be modified by user, when, in fact, we are not expecting users to modify those files here, but to create them in the projects directory and during our drinkable conversation, we came to the idea that it may be worth to move these files under the Lib directory, similar as the CIT Empire store. So it follows the common pattern over the key club and I supplied. The link to the issue and discussion happens and I just invite everyone to chime in and step in.

B

So if we move those files, users will not be able to change, but.

H

Yeah because we'd not expecting them to change them as far as I know, so the common metrics or the system metrics are those these default dashboards, and if user wants to modify them either they duplicate them to the project or create the custom metrics.

H

We are which are stored in the database, but they never touch the files and, in fact, if you want them to allow them to modify those templates files, it will require a lot more a lot of more work, because that, in that case, such files could get get unsynchronized over multiple physical machines when the each lab could be run in the distributed environment.

H

From my understanding, I don't see any risk, but if anyone does break it up, if anyone see any risk, maybe I'm just missing something. I'll be happy to get some more feedback on that before making such a change.

H

That's all for me next to beat amigo points but she's, not in the call I, don't know who she'll verbalize this.

A

Sure I can talk, so we can look at the link. There think this is just a kind of a discussion topic and in me, I was looking for some feedback. So I don't have a lot of background. I, don't know if anyone else on the call does yeah.

F

I think this is about how absolute and relative time ranges work and the date I figure that we have in the dashboard. It seems to be getting some traction, as in other teams, and also interested so we're planning to make it a genetic component. So for that, we are also thinking about how we can make these components more compatible with scenarios where we will have to auto refresh or just refresh so yeah I think this is. This issue is about how we can deal with relative and absolute time. Ranges within that component.

A

A

They also have topics they work to cover. Do.

B

We want to go over, so I saw the Planning Board for 12.75 issues.

B

Just quickly go over them, I think.

A

We should go over them yet I think they weren't move because they're in workflow verification, I think that's a special case that mm-hmm check cuz if they're, that presumes they're already merged and we're just making sure they got into 12:7. But I could be wrong on those but yeah. It's probably worth just reviewing those I couldn't yeah.

B

I see one of the old flows in education, I.

A

Can share here. A second was gonna. Try to hide some of the security issues.

A

Let's do this: okay, so yup there's these five issues, so maybe we could just walk through them. So the first one, let's uh toggle for the self monitoring project.

I

um So that's that's being merged everything from ten on mekin and we found a few bugs, so they are working to fix those bugs the bugs have separate issues open for them. So actually we could probably close this one.

C

Under a future fight.

B

The back someone scheduled to the next iterations or the critical path, of course, I.

I

Think two of them are critical, bugs which we already have a markup for I. Think both of them I in maintain a review so yeah.

A

Okay, the next one or a snowplow events or custom events, yeah.

F

So that one has been merged as well, I'm still waiting for it to be deployed its to the notification and issue after that, if also mine, that is the same situation, is a dark ticket. It's been merged and I'm waiting for the in deployed just a very fun.

A

Okay, are they gonna get into the 12:7, deploy.

F

I have been worse over a week ago: I don't know if I'm using something or the process change but I don't see the the comment that gets added to the merged, except that it says it's deployed to no canopy or production or develop. I.

F

I

F

There are 15.

I

Which was the last Wednesday was the last last day that you know changes went into then.

C

You go ahead and try to test it like for the data, because solar 7 is released there.

F

Yeah Marshall on 6th of January or 50, probably just try it on the production and see.

A

Yeah I think the process changed a little. They don't. The bot doesn't add a comment to the mr anymore, but it does post it in like where the pipelines are here. We can see it on my screen, so yeah.

B

A

Is the one of the snowplow ones? Was, you know prod week ago, canary staging so it has actual dates. So it looks like those those were merged, so yeah be worth testing him out again: okay, yeah great if they made it into 12. Seven that'd be good. Okay, all right this next one I think this is the one that ya hoarder was working on the mug that he opened and also fixed, so I'm, not exactly sure.

A

Oh I could follow up on this one and then the last one. This was one of the bugs right around the self-monitoring project.

I

A

I

This one actually made it into 12.7, so I'll close this as well. Okay,.

B

Since I was not here for 28 planning, you know anything that I think we should sell Fisk's and, like the amount of issues I see, we have 37 issues for 12.8.

B

C

I jump in real, quick I didn't know. I forgot to mention this up in the agenda, but I do want to mention. We reorganized the way the board is so the deliverables are on top and ordered based on priority, as best as we can to copy the planning issues, and then we have the topic concept of filler issues right below this part, which is basically non deliverables, so things that the team can pick up after they're done with deliverables or doubles our block and then below on.

G

The way at the.

C

Bottom, we have a verification section, so we can, you know, pull things down as things get verified, so it's a little bit easier to see the list. So, even though it's as 36, it's not like 36 things have to be done. This release. It's really everything that it's a deliverable that we're saying.

I

We're committed to so.

C

Hopefully that helps clean things up a bit.

B

Okay, thank you for that.

B

When I said, we want to donate mine as well. Yeah.

C

I think I think we should try to do that, trying to keep the deliverables to like maybe 20 or something like a lower number and hopefully that's what we decide in the fighting issue.

D

C

Know we have a couple minutes left yeah I would say: do you want to mention about some multiple metrics alert that one seemed kind of recorded, I'm, not sure yeah.

E

So, basically, for the support for multiple alerts, I well with the help of Brian, to encounter a race condition on the back end, which pretty much destroys the Prometheus server. It just remains in a constant loop where it will remain updating and it will not update whatever changes to the alerts that you have made and on the front end. At the very least, I found out that this is becoming more and more frequent to the point that is blocking this particular feature from being released.

E

If you start adding or deleting metrics at a reasonable pace, I mean alerts, it will completely block the primitive server. So I'd say we cannot ship this, because even if we have the front already, people could encounter this race condition and pretty much.

E

You know, destroy the workflow.

B

So, what's the of.

E

The solution for that one Brian specified that we need better a error handling nor, if I recall correctly I, don't know I'm sure Brian asana Kolk, you might have better.

C

So so there was initiatives created and trying to advocate for it to be a prioritize for 12.8 deliverables, unblock it, but I'm not necessarily sure whether there was a clear path to fixing it.

B

So do you think maybe we should like move it below I'm talking about the fondant support, multiple metrics, that I tried it's like this, the second from the top you want to that, put it down and the priority list, because this tree will blocked from.

C

Well, basically, if we don't part as a fix, that's not gonna ship. This release.

B

You have a link for this week's.

C

Yeah Sharon's, like the let me post on the zoom I, mean I. Personally. First talk I.

A

D

It in our agenda as well at the bottom there, okay, so it's need to prioritize. Do any of.

B

You have any ste patient publish time. Will it take to implement this.

E

There you know I, since this is all back in related I'm, not sure. What's the what could be the timeframe for this.

F

Also for the bagging issue looks like Peter posted a comment that maybe a possible Warcraft I wonder if we should try that and see we can get around this for now until we have a full fix enough.

E

Chuckle war to do bachelor creations.

C

Do we have capacity in the backend to assign someone to this maybe depart as existing back in billable, since this seems more important based on my perception.

A

Okay, I could look through the list of the deliverables and see if there's good one, that we could push down a little bit to prioritize this one. Instead.

A

So yeah I can take a look at that.

A

B

At that, and let me know, met what you thought: I can't be the one, but we can do.

D

D

Yeah, thank you very.

A

B

J

Issues raised well I, just completed the sentence in the document related to using gitlab UI in the application. I noticed that for some elements, namely what I'm working on right now, it's a drop-down that has to to fill the container that it is in so varying the width of a drop down. This is not.

J

This behavior is not defined in gitlab UI, but the issue I'm working in presumes that we have one so fixing the issues like this implies also making an issue in G flat, UI and fixing that one and using the solution in the application or should I make it on the spot.

J

J

This is my concern right now, because I saw that usage of gitlab UI is recommended over making.

J

Quick fixes in the user interface in the application.

F

Yeah, usually, the recommendation is to not have custom styles, especially for components that come from bit, laggy right and if the fix seems like it should be in github UI, then yeah it should go to bit LabVIEW and then we just consume that in jet laughs, so yeah I, think I, think and also the principle strip should are usually from the gate lab designed, pyjamas design system.

F

So if there's a component that doesn't match the design system so that it should be fixed until our key or a and and then it will automatically destruct on the job, I want that whole case in person. Yeah.

J

That's what I that was my opinion also and I I wanted a confirmation for this. Thank you.

C

You yeah I'll, just add what I was typing I think that a lot of it depends on the effort and where we are in the milestone right. Our highest priority is to ship features or one of our highest priorities, so we also want to be consistent at the same time. So we try to balance that, like, ideally, it's in good lab UI, but I'm. Also, okay, if we could do as a follow-up.

J

A

All right great, um is there anything else that people would like to cover.

A

Once all right sounds good: well, thanks everyone! So it's good to have you back and yeah, we'll talk to everyone soon, I'll get the rest of your day.