GitLab Support Team, 29 May 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Support Release Review Party - 13.0

Description

No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).

A

B

A

Thank your. The checks take your own, so we're waiting on you yeah.

C

I'm, just waiting for Ariane to come in okay, I believe we now have everyone ready. So shall we start all right today is the 29th of May, 20 20, and our release review today. We'll focus on 30 now believe this is to me the first major release review and for anyone wondering we have started our release preview party with 12.2. So this is going to be an exciting release, as we now have an improve it. Early surveys, giddily cluster and auto deployment of gitlab to AWS, GCSE, application, security, testing improvements and so on.

C

We will look into the details later, but for the first in the list, I think Andrew will start with us. Do you have a good chat on what is a major release for support? Andrew yeah.

D

So I'm trying not to talk too much about this, but I just trying to make everyone on the same page about what we mean by major release. Again, it's not so much about feature from from our perspective. Is it more about breaking like backward incompatible changes I? Think waiting is gonna, talk more about the definition of the application and stuff like that. So so that's really what major release means for us.

D

D

Yeah, the other thing is uh I. Think if you look at the notes, let me see if I can just share my screen quickly.

D

What do you see use your browser? The supporter is not probably not am I sharing the wrong screen. No you're sharing the release, notes.

D

There's no sign here, what's going on all right anyway, I think one of the things that I wanted to highlight is this part, the extra great stuff so.

D

This was not a thing before version 12 and if people around during this period, there are a number of upgrades from the customer when they skip version 2.0, and there are some issues with that. So does anybody know that we actually stop people upgrading in South manage if they're trying to go from twelve point ten to thirteen point, one straight away: I can't remember if there is a.

D

Something that stopped there from how many bus I think.

E

I know why I would have done this because of that issue. We had at version 12 when people jumped versions and it removed the migrations that were running from the major version upgrade right.

D

So yeah I probably need to like look into this further myself, but I guess the point was when we replaced 4.0, we the release. No, it's saying that you should go to a point: zero first, but obviously customer probably missed that, so they upgrade straight away from from eleven point eleven to twelve point two three or something, and there is something that's breaking on the back end I'm, hoping that by now we actually have something in omnibus that stops us from doing that.

D

When they're trying to upgrade to thirteen point one they'll say: hey, you need to go to 13.0. I'm, not exactly sure is the case, so something to look into.

D

Yeah, so there's also a link here about major release fabrication make it release note, so you can have read about that one other thing about for us in support. What does it mean? We have this note about what version that we support. We basically to support two major versions.

D

So, as of now, we support it in 11, 12 and 13. We're not supporting 10 anymore. Is that right? Is that my understanding right?

D

It should be the case, then, okay, so that's another thing that what it means for us in support specifically- and if you want to like read background discussion about inter release, you can go to that there's a lot of discussion and the executive level, but why are we doing this and things like that? So yeah? That's, that's all I wanna talk about any notes questions. Otherwise we can the one.

F

Give me the process of killing people of 10 I believe in their own 10.

D

We just basically tell them so I think when, when the definition, if we don't provide them, support it's more about, if they have a bug and they're like hey, this does not work and I'm running ten point. Something then we'll say: actually we don't support them anymore. Please upgrade to later version. This is the upgrade path going upgrade and then we can take a look into your problem.

D

So it's more like that. That's my understanding. Anyone have any thoughts about that.

D

Yeah there are some tricky part, what happened say: they're running version, 10 and they're trying to update to version 11, and they have an error when they're trying to upgrade how we can support that.

F

Should I think.

D

So too, but again it's a case-by-case basis. Right I mean there was some tickets, where the customer is trying to upgrade from version 8. For example, right you even click months ago version 8 is like so old already and there's some there facing so many issues. Maybe what we can told them like look the export the project and then import it to like a new version of give up. So there are again it's a case of Cape Isis, anything so yeah. That's another example that I remember good question.

D

C

Know I think we should put an mr2 update the current versions in the statement of support that you link there Andrew.

D

Yeah it says his, for example, say: I became II can do that as well. Yeah.

D

Anyway, I'll share my screen and alright.

C

Pass this on to blaming.

B

Application, because this was clarified very recently- and we did on this page over here, but the key thing is, you know: maybe we move at a specified point in the future, so this actually came up because at some point our documentation. We stated at the CIA syntax I'm.

B

D

You're going out, sir oh yeah.

B

The CI only an exact syntax was a candidate for the application, but you know we didn't actually save on staff occator or when you'll be removed I'm. So there was a number of user confusion and we did receive a number of tickets as well where people are like yeah. You know if this is being deprecated.

B

I need to change everything to the new rule, syntax, but at the same time the roof syntax doesn't cover all the different use cases that only an accepted, so the product team went and I she looked at it, and then they discovered that yeah that shouldn't be in the documentation at all, because it isn't actually deprecated. We shouldn't say that we are pretty deprecating something so at the same time for support, because we use the dog so much I.

B

Think if we see any application notices in a dog spit back to the application, they all remove a day. I think we should update the documentation, try to add them all. We discover that you know it's deprecated or if.

G

B

Say it's deprecated, but there's no renewal date and you can't find any information about when you was deprecated or when you'll be removed. We should raise an issue and then get the product team to look at it.

B

So moving on to some of the actually before I move on, you have any questions about the definition of the application.

D

Yeah that example, that you mentioned, but only an exam I, have some own personal discussion of that, but I think that's outside of this. The scope of this particular chat but yeah. That's that's good to know. Thanks.

B

Yeah I mean, if you have time at the end of this I mean we can have that discussion. You might be useful, but yeah, moving on I've selected three applications which have near-term. Why impact meaning that we'll probably get customer tickets on these items? So the first is that unique on is deprecated and Puma is now that if our app server for new installations I believe for existing installations, nothing will change.

B

But yes, the biggest one is that Pumas maximum request? Duration is set at 60 seconds and is not configurable. So this has implications for longer running requests. We often troubleshoot issues where some customers need to increase the unique on worker timeout, because typically they have dependent get any requests that take much longer than the default 60 seconds to complete so donee work around. This is to revert to unicorn and wheel as I. He created a issue for this over here and we were actually done.

B

The video was true of Puma time now and what Puma does in treads as opposed to what in Econ does, which is spawning additional worker worker trends, so this would be removed in a year in jail at 14, I mean Union unicorn over remove at 14 right yeah, unique, only removing deal at 14 yeah, so I'll go through all these, and then we can have questions at the end. I think that'd be easier. So the other thing the next thing is legacy. Storage for projects is deprecated.

B

Now hash story should be used by default for new projects.

B

We have actually moved to using hash storage by default for a very long time, I think for a little year now, but previously administrators, you had an option to keep projects on legacy storage, but now that has been removed and I think the key thing for us is that in 13.2, all projects to in legacy story should be automatically migrated. A background migration I'm, not sure what the mechanism for this is, but I predict.

B

This appear potential ticket generator, especially for you know, customers who are upgrading from old versions like 11 or yeah stuff, like that. So for those who don't know legacy storage is basically the path to reposition. This basically reflects the project's restful, hash storage. We actually store them using a hash of the project ID, so this actually makes things much easier to manage from application perspective.

B

We know no longer need to synchronize the file system with database changes, and it prevents problems that you know if you create a new project with the same path as a previous. The later project, if you had legacy storage that might have created some issues. So again, this would be legacy. Story should be finally removed in Yale at 14 in a year in the last.

D

B

Gap sorry go ahead. I was.

D

Kind of imagine there would be like some if they have issues with their underlying file system and if they need to move from legacy storage and have storage running in the background it might have like some perform an impact you reckon, but this is this is likely right, like the forced move is later yeah.

B

The first move is in thirteen point two, so I'm not sure what mechanism there would be, but I think it's probably just background migration, which is actually basically just I, mean despite the name background migration. It really is just a psychic job that stuff in the background, so I'm not sure what the exact mechanism is, but I mean if, if they are detect that fall, Commission issues, yeah they're very well projects disappear, and then we have to troubleshoot those yeah.

D

You know it's our future problem. Thirteen point is.

B

An FS, it's deprecated, everything you'll, get questions about this because many of our customers do you. Do you use NFS, so I'll cover the reasons for deprecation first, so the reason why it's been deprecated, because NFS has been known to cause data loss, especially when there's high right pressure from multiple clients. So realistically this would mean you are running multiple attorney servers and you're sharing the file system across them using an NFS.

B

Apparently, the reason for this is because NFS, even in the best-case scenario, is eventually consistent, not immediately consistent. So you can iron results where you actually push it commits, and then the repository actually thinks that yes still hit instead of the current hit. So basically you get your kid. History is actually overwritten unintentionally, so the replacement for highly available environments, you insure planning to run multiple attorney service, is actually keeping it cluster, formerly known as little eh-eh, so support for NFS we finally removed in Gaelic 14.

B

However, it's unclear please remove us or effect non-ag environment. So a lot of our customers, basic use, mount get storage, Shasta NFS mount and they do this I guess for durability. Reason so that they can actually let the VM be responsible. So I'm not sure if this section affects that kind of use case. So I will follow up on this one yeah. So that's it for my selected applications. Any questions on any of these I.

D

Again, the NFS park right I actually have a little bit notes about this. That I wanted to kind of talk about it a little bit, but maybe I just bring it up here.

D

If you go scroll down a little bit way on the notes that I wrote, I highlight it done right, so one other thing that I was noting is that if you go to prefect support at the document there, it still mentioned that prefect is experimental at this time and we use prefect for so I believe that prefect. This is like a built-in.

D

What, if I'll get router thing right, so we built this ourselves and it says, get the support experimental and we do not recommend using it production this year. I, don't know how that works with NFS being deprecated.

B

Okay, so in this case, what we are doing is that we are announcing the NFS is deprecated, meaning that we will rewrite it in the future. So the reason we are announcing it now and then basically actually removing it onion for dinner, so that users and our customers, who have a year to sort of plan Mike.

B

So from my understanding, many customers I have actually been waiting for Italy cluster for a long time now, and the Catelli team.

G

B

Working very closely of them to sort of move them over, so that.

G

B

Before we actually reach the removal date, Kotani cluster will be a mature product or mature enough product to be used in production. Okay,.

D

Okay, yeah, that makes sense now. Yeah I was reading that and I'm like what's going on here and that's why I put an opener on the bottom that allows a big customer based on system performance looking into this, so, okay, thanks yeah, that makes sense now. Yeah.

B

Baby, it's so it's.

D

Not really production ready right now, but we are going to deprecate it, but as of 13.0 and if I still be like recommended thing.

B

Right, basically, sorry I saw 40 nodes of the team as of.

D

B

D

B

C

Least, 30 well.

G

B

Wouldn t recommend it so much as there's no other alternative yeah.

D

Yeah graduation, that's good.

B

But yeah, if you, if you suppose it I mean even if we moved to remover in 14 I mean if customers are still on 12 or 13 I mean it still sort of in support. Tell me you still have to support that. I mean it's a bit confusing I! Guess by that design is there's a difference between product remover and support remover right.

A

B

To think about this,.

A

Is basically letting all of our customers know get ready? You have a year to prepare for this. Some customers will have a change management cycle that could be multiple months, so they may need to schedule something six months ahead of time before they can do any work, so this is making sure we're giving customers as much notice as we can give them of this is this?

A

Is the change we're going to be removing support for innovation, have a year to get your environments ready for it and for those customers that do require some kind of replacement? This is what it's going to be start playing with. It's not production ready yet, but you can start to understand how this needs to be incorporated into your future architecture.

A

So, yes, we're still supporting it, especially if customers are an older version of gitlab like yes, we're still going to be seeing tickets about this also important to note, even though it is going to be fully deprecated in 14, because we support up to two major versions back, we are going to be seen tickets and we will be seeing issues with NFS for at least a couple of years after that, so this doesn't mean that, as of now ignore NFS, this just means hey. Let customers know this is something that they need to prepare for.

A

Think of it in those terms, rather than what it actually means for us right now. Today, yeah.

D

That makes sense thanks for the clarification sure.

B

Prefect Oh talk a bit about it as well. So basically, beef is a reverse proxy for Kotori and why it does is that erupts it routes read and write to the master or primary replica of Kittery. That's actually work being done right now to enable reads from replicas, but not right, right to always go to the primary. So it's a bit like a favor over plus reverse proxy.

B

At the same time, how it works is that it also has a separate Postgres database, which it uses to record I, believe practical information and other things like that and somewhere out there there's an issue where they're actually investigating the possibility of even a single note omnibus, install using Prefect. So they may actually implement prefect for all use cases, not just high availability, but person.

G

B

Still very much preliminary, but I think the message is basically yeah. We should learn about it, because people can come to this more and more yeah.

D

That was another thing that I was interesting for me from that particular diagram was there's another posters database that we might hate to kind of manage yeah.

B

We do so the current the current issues, with integrating it with omnibus from man and standing is that we Omnibus currently is not able to set up that separate prefect tracking database for Postgres they're working on it right now. I saw an issue about it, but yeah yeah. We will probably look at normal.

G

Tickets about it sometime.

B

The future, so it quickly start thinking about it. I was.

D

About to say that I'm sure we gonna talk about it again in the future.

B

H

Thinking in terms of like overall world view, if we have like I, think we can extract the list of customers from genders who have asked any questions about NFS and we can start sending notification that ok, if you are thinking, then we can plan something out. So that would be that would make our life much more easier, because this is this might get skipped from the eyes of the customer that ok, the you are using, that we are in the process of deprecating, get an H NFS in in the future. So maybe.

A

C

A great idea: that's.

A

Probably something that's more per custom success, though, because they are the ones that ultimately have that relationship have a think through the best way of doing it. Like I, like the idea of an issue and then publicizing the issue as a heads up, I, don't know if that's the best way it appeals to me, but maybe that's something that we can talk about offline Ariane's.

D

Can actually add something to that we, the customer, should receive a weekly or monthly newsletter so like as usual they're like hey new version, hey something happened with youth level or whatever it. So maybe there's something there. That's already this part of the system that we're doing but yeah I like the product, if necessary, on.

C

Okay, thank you for that waving. Any other questions that you guys ask Amy going once twice all right. I'll put it into my part here, I'm just going to share my screen.

C

Okay- and you may see my release- notes right now. Sorry.

C

All right, I'm going to touch base again on the tip equation that Wei Ming has inform us earlier so from a github.com service perspective there, where there's a lot of tickets that we have received, that involves CI CD. So one of the widely used practice now is the deployment of pipeline using only an accept. So, as you all initially know, only an accept was initially planned for the creation and format. What format from what I understand? Is that the the docks we have here on?

C

Only an except didn't have this note previously, so a lot of the customers will actually confuse of this notice because it was previously set as they're going to the brand deprecated and it was not set as a option that they prefer rules from a product perspective. So what happened was that customers on get lab comm migrated right away to rules, and you know try to explore and discover options on how to how to apply it to through their the project.

C

Saint Rose, it's not fully, let's say, mature or develop some of the like logics that they wanted to implement. It's not available. So much of the customers are still relying on only an excel. So one of the key factor that I'd like to highlight here is that the use of only an accept is only applicable to both routes and only accept if you are using a custom, get lab CI ml file.

C

So if you, if you see a customer using a a template like a SAS or that kind of CI which I will demonstrate here, if I try to edit this one- let's say I put in the only only section here to include say just tags and schedules. Maybe.

C

Now you will see a good website configuration invalid, which, which contains that the it's? It's only accepting only if you see that so there's a lot of pipeline test has been broken, which I have seen in get Lancome. That is not that is no longer working because they have combined only and except together with the rules when so, when they're, using the template obsess here or any other auto the auto depth APNIC template, they shouldn't include the only tags. Okay, so that's the only thing that I wanted to emphasize.

C

Any questions on that.

D

Yes, so I think I think to be more specific on that so rules and only accept our exclusive mich mutually exclusive. Is that right word, so you can't.

G

You can only just one or the other.

D

C

You can use both for a custom, get lab CI yeah no file really.

D

C

That's the that's the thing that they were working on. The only problem is that they're, their validations are happening in the dissipation.

C

So if the tub configuration templates are included, our validator will then prompt an error that you can only use rules. Yes,.

D

So that there was, there was actually what I was about to say. It is I need to look into this further. Maybe something changed what recently, but my understanding previously was only an accept and rules are mutually exclusive, so you so the customer can only do one or the other. Maybe there's something different now, but yet this pretty sorry.

C

What about the confusion with the customer as well? The reason right, we've had kind of like frustration over this change, yeah.

D

I think I think that's that's the thing like so the yeah. The story is that if, if there's a Auto DevOps right back, then when they want to configure all the DevOps the customize of the DevOps, then they use only and accept syntax to do something. But now the template is using rules. So they have these rules to customize and that's how it break a lot of the customer stuff. Isn't it yeah.

C

And from one of the issues that I've read, I think one way or the other in the future, this will be fully duplicated and rules is the only you know, syntax that will be used. So only a next step will be it's like a slow transition to advise customers in a sense that if you ever come into a tickets such as like this, you know it's best to also advise them kind of like that, we're going to deprecated in the beecher. We were just preparing rules over now.

C

For the meantime and the next one that I wanted to touch is about the email notification for anon sign-ins. This is specifically for get lob, comma support people.

C

The reason I raise this one is that it appears that the product theme is more of like looking on how they can, let's say polish, the way the youth love.com logins is working, so I can imagine this can potentially get some tickets requesting some Cabana logs and so so forth, like more on the legal legal part of asking who has log into my account so I believe if you're not very familiar with Cabana and other logging tools, we do have a support workflow about that to touch on that, because the other day there was a distillate asking for why you see getting some from self login from an unknown location.

C

So in a sense of you know, if you're not really familiar with your vanity, it's one of the things that we can focus on as well, and the next one is the added admin personation events I have tested this on get comm from the feature itself. It says that this should appear in the admin logging under audit events, but it appears that after I impersonated an account, it does not appear to show in the public project level settings on github.com.

C

Yet so for now, there's nothing to you know be worried about of impersonating accounts, because this is what get that becomes. What engineers usually does they kind of like personal accounts, to replicate what the customer is facing. So are you trying to look into this and if you look into the issue, it says that it's a lot more complicated on the compliance side or get lab comm. Since it's as admin who's, you know everyone will be I, guess I'm, you know, there's no specific facility on the way this logging is set.

C

It's like it's only showing us impersonated by an administrator and it's not specific to the user yet whose specific administered so I think that's the part where this is not being compliant yet. So this is something to watch one and for the next feature that I'd like to highlight. Is the group level push rules, so this was just recently deployed, I think very fresh just yesterday. So if you're looking for the push rules, it's actually showing on a subgroup group menu here, I.

C

Have not seen much tickets really related to push rules yet, but I will think and imagine that there are a lot of projects that are being set already with the push rules. So if one of the customers are asking why the push rules on the group level is not working, this is because of the project level report repository push rules here, so if they have a push rules on a project level, it actually overrides the group push rules from my understanding of the DOS: that's how they do it.

C

And also when working on github.com, please ensure that the pushers projects, a project or group subscription, must be started on bronze, high bronze plans and higher. So any other questions related to this features.

D

Specifically, but they kept group level push rules for self-managed, it's actually currently still behind a feature flag and it's currently disabled by default. That's just additional thing, so I'm, assuming is enabled on gala calm, but in sub manage is disabled by default. At.

A

This is during a quick time check. We have 24 minutes left.

G

Thank you. I'll try to share my screen. I.

G

Think unity and share your screen. First.

G

So next feature is the version snippet, so before snippet is just a snippet without version control. This will allow. When you create a snippet, it will create a separate or repository for video for your project or your personal snippet, and by default it will has a master branch where, where you cannot delete or make new branch. So if I go to use the pet here,.

G

So you, you can also have a choice to highlight your snippet as well, so I'll try to tell skin labs yeah.

G

So once great, you have a choice to clone a snippet like a normal repository and I'll. Try to edit this to see if there's different version.

G

So we'll clone with the d-pad number.

G

Let me do get lock. You see the full sneak history here so currently, all the previous snippet will convert to sleeper repository with our initial commit as the commit message.

G

And currently there's a limitation or the snippet binary file is not support and also get tacked. It's not support and it has a maximum size of 50 megabyte fortify. I.

H

Would have loved back at 14. Snippets should have a history button where I can simply can see all the history yeah.

G

D

I get there eventually yeah.

G

D

Is just the first one? That's.

E

In the past, when we had snippets to give the customers would lose track of which was the correct one yeah, because people would modify them and there was no version control and so we'd have all these different. If it's going up but yeah really cool.

D

Another note about that version: snippets was uh so.

D

This in self-manage perspective, when they upgrade to 13.10 there will be a background migration that will convert all the snippets into a version snippets and there is associated write tasks for it. So just bear in mind. There is some maintenance tool behind it. Then, if something came up, I I doubt there will have much tickets about this, but it's good to know anyway.

G

All right next stop exciting duck theme, web Heidi, so yeah this simply, you just go to your profile, preference and check the top seat, and this is how it will look. It will only show on the web if you open the web IDE so on your normal lap. Browsing through gitlab you'll see the normal screen so yeah. This is society. It's invite me to use my ID anymore.

G

D

I can I say a fun three here related to distorting. If you see, if you see the issue and look at the one that voted the most, it's actually deep, left dark mode so like the whole, get lap in that mode. So is that the highest folded feature that was asked by the community so yeah anyway, yeah? That is a fun trivia, but anyway,.

G

Next, we have our animal cruelty, 4-inch protection, so now you can set your default range protection like fully protect developer. Cannot push meant maintainer can so these are the option available, but after testing it for a little bit, I see that it only affect like. If a new project is created within the group, it will not overwrite the project protected branch, so that is quite weird: I'm, not sure your faith, our intended behavior, so you'll need to look more into this.

D

Yeah I was read a little bit not about this group level. Sorry, the group level default protection branch. It sounds like a pretty simple feature, but there is a lot of complexity in terms of it create a lot of weird user experience when a customer create a project- and this is kind of one of the things that we're trying to sort of fix from the wider experience of using.

D

So we do get a lot of tickets, not a lot some tickets in the past when they like, hey I, create a project, but I cannot do ABC and things like this yeah. That's that's kind of like the complexity around it. So it's good that we started doing this because it may get it give me system, administration, more and more control, but at the same time it is actually more complex. The more option we have the more complicated it is, but yeah, that's just a little notes for me about that part.

D

I'm excited to feature is that but I can see, there's some potential issues. Hopefully the error message will be clear when they cannot do certain things. Program, yeah I know.

G

All right back to YouTube.

C

All right running short of time so I'll, that's the Sun. Sorry.

H

Yep, so let me show my screen.

H

Yes ended up right, Yes No, maybe all right so the 13-point. Oh, we have made the deployment to pcs much more easier earlier. The deployment part was mainly focused on GCP and Google clusters, but the 13-point. Oh, we have made auto, deploy easy s as well. Easy s is in plastic container services, it's a basically in obvious product.

H

If you look at here, all you need to do is set certain variables with your eks cluster and you should be good to go of it using the deployment part or the auto divorce part or fit lab. What I felt about it is. There is some information missing on the documentation part regarding the auto layout section. So whenever user comes in on deploying to EWS, he usually preferred cloud deployment documentation, but it doesn't talk much about how auto the auto devops can be helpful in this case.

H

So it says that okay, you need to set certain variables, but it doesn't talk much about how auto devops is going to be used in this case or how they can use auto devoxx, although this feature totally is meant for a flavor, but.

G

H

This is just mine in standing, but overall we do now support AWS deployments, which was previously not that great of a user experience. But but this I think we have made a little bit easier for them to deploy on much clouds.

D

Any question so I think I, don't know I, don't know, however, actually past this part this year. But my understanding is that with auto deploy to ECS, that means that to use a complete, auto DevOps and when they don't need to attach a kubernetes cluster anymore. So that's another sort of extra. If my understanding is correct, so.

G

If there's no thing Canaris, then.

D

They can do that with this right, because back then that you need to have kubernetes to dr. source if.

H

It is look at this far there. They mentioned that if you have a kubernetes cluster tied to your project, so it will refer kubernetes deployment rather than it'll, be right, plumbing right. So this is what they have I believe that says that this is the coming feature that you can select which cluster you want to deploy your app, but for now it prefers kubernetes over AWS yeah.

D

I guess I was sort of mentioning that. Just because say, I would like to use auto DevOps, but I wouldn't I don't want to use kubernetes back then then it's not an option. Right I mean you still can do the testing and so on, but I can't to deployment without kubernetes. Now I can but I mean the options are limited because you have to do it is yes, but yeah.

D

E

Kubernetes, it's just I'm.

D

E

D

E

Yeah yeah I guess.

D

Forget what I was saying all.

H

Right so moving to next now we have their application firewall as I am integration, so SIEM integration is nothing but security. Any information event management, where we can check the Application Firewall logs directly into our into the gate, lab instance, or we can say, get up. We can configure it. So this is another step towards making strong word different stage. So this is pretty cool thing we internally use fluent, be that can use the open source. That's an open source app through which represent data collection and basically, that's a data collector.

H

What I'm missed in this part is our documentation doesn't mention how this is going to be implemented or how this looks when it gets implemented, because I didn't got a chance to test it myself, so they simply says that ok, bu, slew nd, all you need to do is configure it and that's it and you should be seeing the data into 20 apps, so I, don't know how that's going to look like, but I assume they can see all the application threads for any kind of vulnerabilities right inside gate lab.

H

That is what my understanding is.

H

Next deprecations, we have removed the log view, admin dashboard, so people who don't have the access to the console, they won't be able to see any logs. So first they need to check their the Clapp local admin to get the logs earlier. We have the option to just go to the dashboard look at the recent logs, but now we have, we have removed it completely. I.

D

Just I just noticed this part: ok, if that's a big one, actually.

A

I'm gonna do a quick time check again. We have 10 minutes left. There's.

G

A

The this look quite a lot on that on the docks. I, don't know if you want to do a condensed version, or if people want to change over another follow-up session, just because there's so much ground to cover with it being a major release. Don't know what you wanna do.

D

Oh so the next notes that I have here again this is like my usual personal note, for anyone who doesn't understand what I'm trying to do here. This is my personal note that I write for myself when I'm, going through the release, notes and I put it here, because I'm gonna need to write it somewhere anyway, so might as well just put it here for people to share and I'm just gonna go to a few of the highlights. So I highlighted with yellow on the notes. I'll share my screen quickly.

D

It shouldn't take too long, Shawn, so I'm hoping I can finish this. You see my suena the release, notes. Ok, cool I shouldn't take too long because there's so much stock in here, but I guess the like it once the interesting point get lab faster ways already talked about it. So we'll skip that part this one. Thank you summary of terraform plan. I, don't know if people notice this part and there's a lot of like chatter, about this on the Internet I've.

D

Never actually I haven't actually test this myself, but this actually create a bit of a buzz in you know social, media and stuff, so basically yeah. We can see the terraform state and I'm actually kind of wondering if we can use this for our deaf resources or even the interview platform, and things like that.

D

So yeah have a look later self.

D

Puma so ways already talked about this earlier there's a few other notes that I want to highlight about this, so something unique like if it on the bus and some of the unicorn settings, they will need to manually migrate.

D

The settings right, so the unicorn worker killer is probably the most widely used configuration so I'm just expecting there's something around this ticket wise so be remember about this and also Puma with NFS back storage and rub enable so this has got to do with the historical performance issue in NFS, back storage and apparently Puma with multi-threading processes have issues with this, so they they need to consider running, Puma a single thread and they run Pumas a single thread. That's kind of unicorn anyway. Right so anyway, have read more about this.

D

A way put a link there about two, more particular so yeah. This is an another notes about the Puma, the changes from unicorn to Puma. Another thing was about charts, so I personally have never installed get up with charts, but there is also this note about the config map entries.

D

Does anyone familiar with this in any case, so yeah boom I still default? As you know, omnibus normal me, bus, the chart is changed from unicorn to web service, so it's like a general web service and again if they want to configure Puma and unicorn outside the default, we used to do config map and then now it's been removed in favor environment variables.

D

These will break some customer charts ugly. So keep that in mind, if you see any things related to this.

C

Curious about the change to Puma, because previously uh from what I've seen in self-managed, they could be used to like provide recommended server configuration with the change to. How does this affect customers? Are they moving more into more like lower my memory kind of server config, and do you, like optimize I, mean lower than the configuration requirement for gift lab and so forth? I.

D

Don't know if we actually even go towards that I haven't look into that further, but you're right. The idea is that Puma have a lower memory consumption. Does that means they can put more user into it. That's the question, I'm, not sure I, don't think we look into that at all. At this point,.

D

I, don't even know that the reference architecture, the the 10 K 2 K and things like that they were doing performance, testing and they've done that in the past, I, don't know if they're, using puma or unicorn for the test to begin with, because they're doing like a proof right. Okay, this is a 10k architecture, and this 10k architecture will be tested with the script that like generate bunch of requests and so on, to prove that it's actually able to handle the traffic for X amount of users.

D

I, don't know if they actually using a unicorn or Puma to test those. That's actually a good question, not sure.

D

No, it's questions comments. We can talk about it more later and I'm sure we gonna talk about it more in the next few released anyway. Let's move on yeah. This is what wrote an attention about group level and yeah I was giving an example about why this is this weird ux complexity. You can have a chat about this with me at some point.

D

New comparison mode for requests again, this is a it looks like a little change, but we do have a few tickets about this in the past. This has got to do with the way we handle this in merge, request the difference between triple dot, give and double dot. This get. This have a lit, have read more about this lots of discussion, details and so on. I put a link there to like an epic yeah. Just keep this in mind.

D

If you get a ticket they're asking like hey I, have this Mr D fees we're what's going on this? Probably one of the reason what it does now the customer actually have an option to to give weight the head of the master rather than the branch things like so yeah. Yeah have read more about this. Just keep that in mind anyway. If you want to see your merge request, that's you're eligible for Sean. You look into that this one I, don't know why this is not in the deprecation of don't furnish.

D

You know the scaling so.

D

The syntax is actually completely changed, that is for the off-peak periods.

D

We can talk about this. A bit more I'm personally fell in love with soccer machine I still remember well of where and when not falling off with it. This is what we used in Caleb calm.

D

So this is talking about the way doctor machine work is that it triggers an api to a cloud provider to create a VM and it will be generated and an auto scaling manner. So if you don't use the jobs, then it wouldn't create the machine, but they also have this concept of having a standby machine so that so you don't have to wait until the VM is created to trigger a job. So there is all this configuration around this yeah like I, said this is completely changed.

D

Now the the syntax I would expect it will break sound. The customer contact.

D

Skip in mind, is there.

A

In quick, because we are a few seconds.

D

Away from being.

A

At time very quick chopin's, you can just stop showing your screen for a second Andrew yeah. No, just a quick show of hands just using reactions with zoom. Do we want to shed your another session to go through more features? Considering this is a major release.

A

Would people prefer to not have another session and just catch up on your own I'm, going to say if you want to have another session, arranged, give it a thumbs up if you are happy with where we are, and we kind of leave it at that, do nothing give it say, 15 seconds, so that everyone has a chance to vote and has a chance to display on the screens. So all those that would like to have another session both now I.

D

Feel like we just need to have longer discussion on this, possibly.

A

But let's let's not try and do that this time, because it's yeah.

D

A

Pretty clear that people are kind of happy with where we are so I'm gonna jump in. Thank you very much everyone for your time, those that presented thank you for taking the time to put your thoughts together, definitely worthwhile reading through the other notes that are on the dock. If you have any questions, probably reach out to Andrew, because I think all of them are going to be Andrews, the female wants that kind of hair or noodles that have review yourself absolutely go for it.

A

Do that if people have any ideas of should we change the release review parties for major releases, considering the next one is only a year away like I, don't think it's an urgent thing, but maybe it's something we can think about. Maybe it's something we want to talk about next month. We see how next month's minor release review goes so again thanks, everyone enjoy your respective weekends and I will catch up with everyone at some point next week,.

F