GitLab 12.6 Release Kickoff, 18 Nov 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Monthly Release Kickoff 12.6

Description

Please add feedback here: https://docs.google.com/forms/d/12-QPpvggEsqCvZnDCnuCjqP53joKwjRMPy4PH-Mqp-I/viewform?edit_requested=true Thank you!

A

We're at the top of the hour cassia and you want to restart the recording, or so we just go. It's.

B

All set to go. Thank you. Scott, okay,.

A

Thank you all right. Welcome to the gitlab 12.6 kick-off. My name is Scott Williamson I am the VP of Product. Gitlab and I am pleased to be your emcee today and to represent the great work, that's gonna, be done by all the PM's and designers and testers and and all the teams behind all the work we're doing here.

A

What we're looking at here is a kickoff page. That's auto-generated! It's in our handbook. It's about gitlab comm, slash direction, slash kickoff! You can check this out and each month this will renew.

A

So all the items on here, I'll scroll down a bit are auto-generated when it has a 12.6 label and in our get lab project you can also see there are videos in each section. So, if you're interested in a deeper dive in any one of these stages or categories, you can click on those and watch those. You can also click into each relevant issue. Here, it's organized by section and then stage- and we will run through this as we'll have section leaders present a high-level overview, there's far too much to cover line by line.

A

So we'll give you the broad strokes of what we're gonna do in twelve six, and then you can go down for deeper detail if you wish so keep in mind. We plan ambitiously here. This is our target for 12.6.

A

Please don't view this as a firm commitment. The things that ship in twelve six will probably vary from what we're going to talk about today, but we'll do our best to deliver the very most out of this that we can so now. Let me introduce our speakers. We have Eric brakeman for dev. We have a rego lewinsky for CI CD. We have Kenny Johnston the ops, David de Santo for secure Josh Lambert for enablement, so we'll go in that order. I will stop sharing and we'll pass it over to Eric Eric over to you.

A

C

C

I'm gonna share my screen here and we'll get going okay, so my name is Eric I'm the product director for the Dead section. The dev section consists of the first three stages of the DevOps lifestyle lifecycle, which is managed plan and create and I'm excited, to tell you about some great things coming and get lots of 12:6 in those three stages.

C

So let's go ahead and get going in the manage stage, we're focusing on two big themes: authentication and authorization improvements and compliance enhancements for authentication, we're continuing to invest deeply into sam'l SSO and get lab comm with the focus on quality testing and provider availability. As you can see in this issue that I'm currently sharing we're prioritizing skin support using octa, which is one of the most popular identity providers in the marketplace, for authorization we'll be bringing some enhancements for admins to get lab comm.

C

The first is the ability to list connected user identities for Samuel SSO. If you're a self managed customer. You already have this ability, because users would belong to the instance, so we'll be ensuring that com admins also have a similar experience, because it's a shared environment, we're also making two big improvements to group managed accounts on get lab comm. The first is an admin toggle to allow group owners to disable personal projects for group managed accounts. This will reduce the risk of projects being forked into a personal namespace and then shared.

C

The second is the ability to map existing gitlab comm users to group managed accounts. What this will do is it'll help. Admins reconcile the case where a user from the organization might have previously signed up for a gait lab comm account using that organization email and help reconcile those users back into the group managed account our second theme and manage its compliance enhancements and we're finally excited to be iterating on providing a compliance dashboard at the group level.

C

This will help admins understand, what's going on in there get lab instance and what activities in that instance may or may not be meeting their compliance policies, we're starting with an MVC but we'll iterate over time to make this dashboard much more valuable to our admins, we'll also be providing the ability to list an inventory for all personal access, tokens and SSH credentials in a single view.

C

This will help you answer questions like what type of credentials do I have in my account, who is the user, how its scoped and the last time this key was used, something that is really important as admins enforce access to their get Lab instance and lastly, something that many customers have been asking us for we're going to enforce entropy requirements for new user passwords, we'll be doing this based off of NIST guidance, and we're really excited that this is finally coming and give up 12:6.

C

Moving on to the planned stage in the plant stages, we continue to schedule huge improvements to both portfolio and project management, we'll be focusing on adding burn up charts to our milestones. Page burn up, charts typically give you the work, that's left to do, forces the time left to do it and can typically indicate a little bit better idea of if scope was added to a milestone. We're also planning some milestone, improvements such as the ability to have multiple milestones for an issue or merge request, and then the ability to define those milestone types.

C

This will help you in the instance where you want to say, attach a milestone to a deployment and also a milestone for the sprint that it was delivered in and while we're still talking about milestones, we're planning to add milestones in the roadmap view, which will add much more value to that view than exist today to help you understand how your epics are lining up to your milestone, dates and, lastly, for roadmaps, we're planning on continuing the work to expand epics in the roadmap view, so that you can understand your epic hierarchy.

C

This should allow users to track and manage at the portfolio level much more efficiently than the existing experience. And the last thing to highlight in the plan stage and something I'm personally very excited about, is our continued work on requirements, management in twelve six, we'll be continuing our user research spike around requirements; management in addition to launching our MVC, which is the ability to create a first class requirement right inside of github.

C

Moving on to the create stage, the big theme for 12:6 is diffe improvements. If you're a user of our suggestions feature in a merger quest, it can be hard to understand the difference in the case where that diff is really hard to find so we'll be adding word diff, highlighting to help users more quickly, identify the diff between lines on the suggestions feature and from a performance perspective.

C

We're also going to be working on very nice improvements to diff loading, but first and while we already do progressively load disks, the problem is they're, not interactive and so we're planning on making disks interactive while the rest of that diff progressively loads and, lastly, we're going to be reducing the data sent back for the diff by only sending either the parallel or inline view.

C

Based on your setting, this essentially cuts the amount of data sent back for the diff in half, and the last thing I'd like to highlight for the Deaf section today, isn't a quick improvement to our web IDE, which is an enhancement to automatically staged all changes in the web. Ide we've received some feedback that staging changes in the web. Ide is a little bit confusing, even though it technically follows the normal get pattern which allows you to kind of select which files you want staged, or not.

C

Most people who use the web IDE and are touching multiple files want all of those changes to be staged and committed to their branch, so will be automatically staging all those changes and making some UI improvements so streamlined that workflow and, as always, there's a lot of things. I didn't get a chance to talk about so go ahead and check out those videos that Scott mentioned, and let us know, if anything, is exciting to you in the issues. Thanks so much back to you, Scott all.

A

Right, Eric thanks a lot. Let's move on, don't worry before C ICD.

D

Yes, thank you Scott, so my name is Enrico Winsky and I'm the senior product manager for a relief I'm going to share my screen. So just once again bear with me.

D

Okay, hopefully is the right street, so the CI CD section is comprised from verified package and release and we have mutual goals within this stage, so I'll be kind of jumping around and showing our different things that we want to accomplish.

D

So the first thing that we want to accomplish is a single application for CI CD and we're working to expand our user base by supporting additional package managers. So in this upcoming release, we're going to introduce support for nougat packages, which is this issue right here, which will actually add support for our dotnet users.

D

We're also adding the capability to to add ability to associate feature flags with contextual issues which is in the release section, and what this will actually do is users who are using both gitlab issues and get that feature flags will be able to see the association of a feature flag on an issue or vice-versa.

D

So it's this issue over here. Our next theme is multi-platform support. So in this release, we're happy to announce the windows shared runners beta, so they're on our team has been working really really hard to get this out, and we're super excited that we're finally at the bit at the beta stage. So this is a really really long hall and we're happy to release it.

D

In addition, our CI team will be delivering support for CI jobless clinician outside repositories, which will actually make it possible to configure the project so that the github CI ml file is outside of the repository. The reason for this is for open source projects that are building packages or distributions that do not necessarily have access to specific repo.

D

This is really useful. Another another feature that we're working on is adding UI support to merge requests with deployment context, so this is actually over in this issue. Idea here is when an appointment is triggered. No data. Li data is currently say to the two length of deployment to the merger quest and new deployments will usually overwrite the deployment status of the old merge request when actually the first appointment is the one that's really really interesting. So this feature will actually support this most poor environment and deployments in github and also external deployments.

D

So this is a really big change and one of the most significant features that we're going to be working on in this release is in the in the verify stage and continuous integration, which is child-parent pipelines, Charron, child-parent pipelines, is, will allow a parent pipeline to trigger a child pipeline. This will allow the child pipelines to continue forward without waiting for unrelated jobs to finish, and the configurations will be distributed it to child's pipelines. We believe that this will improve performance and collisions and are really psyched about so diverting this.

D

Our next theme, the theme is speedy, reliable pipeline. So again we have a mix here of all the different stages here. The first one is limit pipeline concurrency using name semaphores, which is I, think our most popular issue, which is avoided both internally and by the community, and this will limit the deployment of any job using the same lock. So if you're using, for example, a deployment for a physical environment or anything that actually needs a limit, this will give you the ability to do it.

D

Okay, so, additionally, look like that. We have two different two new features that are coming up for code coverage and this is actually requires some some showing off of the new UI. So the first one is making code coverage more comprehensive, so the testing team is adding a graph that will show code coverage, changes over time and well. Well, we really understand that the current merge request graphs are are useful.

D

It's really much better to see it over time, and we think this will give a lot of values of the customers and another one is adding a full code coverage report which extends the report that we currently have, but it shows it for the entire code and odd only for a significant code that was just changed, so we're really really excited about these features.

D

Trying to open it up.

D

So, for the sake of time, I'm gonna say sign, but please go ahead and look into the issue to check out the progress themselves and our last thing that we're going to work on in this release is doing powerful things easily. So as a move, so we're natively supporting the hyper cloud deployments.

D

We're going to be adding built-in wrappers for AWS assignments, which can be also checked out under the release stage, and the idea of these appear is that we realized that the plane to the cloud is a super common task and we're going to be the providing base, images that are reliable and work well within the get lab which will make your life and job much easier.

D

In addition to that, we're going to be adding assets to release the other gitlab CI mo file to complete what we started in the previous milestone, allowing users to create releases directly from the git lab CI mo file without having to go through the API. We're also going to be optimizing. The container registered garbage collector collection algorithm in package, which is going to help our enterprise customers will have more than one terabyte of data.

D

Now in the past, this operation could take anywhere between a few hours to even a few days in order to run and complete, and that was depending on the amount of storage you have in. This. Optimization will reduce this to about 4 hours. So we're really really excited about this release. There's a ton more content, but I didn't discuss, but I invite you to go through the videos and through the issues and to officer Kenny.

E

Cool things all right, yeah I'm, if you don't mind, stop being sharing I, will share my screen.

E

And awesome work, working off of themes, I feel like I need to do that for mine all right. My name is Kenny. Johnson I cover the ops section. The ops section is comprised of configure and monitor stages. Our aim in the ops section is to make common operational tasks that developers need to perform easy. So we'll start in the configure stage in the the videos are great. They also have links to the planning issues review all the kind of discussion that the teams had around why we chose these issues specifically in the configure stage.

E

We're going to continue some follow-up work from our in 12.5. We are adding the ability to create an Amazon eks cluster directly in the gitlab UI. So we're going to be doing some follow-up UI improvements from that effort. We're also going to be working on improving the way that you can customize configuration of your gate lab to manage apps again something we started with some kind of base primitives in 12.5, but we'll be adding to you in 12.6 and then also our ongoing efforts to make auto dev ops, smarter, auto, devops, azar kind of defaults.

E

See I see the pipeline that you can automatically have your projects, use we're going to make sure that it doesn't fail in cases where it really shouldn't keep in attempts to run we're also going to be focusing on our server list, architecture, support and moving that from minimal to viable this release.

E

So we're going to be doing that through some use of templates, we're going to bundle some default templates that both integrate a service application, as well as a front-end, so kind of have that combined front-end back-end, using a service architecture which will help developers new to server lists, get started. We're also going to be adding some improved documentation around how to debug your service applications and then, lastly, also giving you the ability to supply a standard domain.

E

So you can utilize the same communities cluster with a key native installation across multiple projects and instances or most multiple projects and developers. In the monitor stage, we are going to be working on continuing our support on on moving our logging experience from minimal to viable.

E

So that will include things like allowing you to you search the logs once we worked in 12.5 to allow you to install a lastest search as a managed app, so we're going to start providing a UI so that you can search those logs in that aggregated fashion, as well as additional support for the kind of log tailing experience that you might want, where you want to see.

E

The live logs come through we're adding the ability to not just drill down into the environment and the pod, but also a specific container and see the logs for that specific container. Well, we're also working to enhance the views that developers can get into their environments. Today, when you look at in environments, the environments tab and you see multiple pods- that that application is deployed to in your kubernetes cluster, you get a series of these kind of bubble chart, but the bubbles are typically well.

E

I present only imply the percent of deployment, so when you're doing progressive deployments, it'll tell you which ones are on new versus old code as it gets deployed, but we're gonna add the ability to visualize different things. Like specific metrics and there, and whether or not that specific pod is healthy on that metric and then lastly, we're going to continue our error tracking integration so we're starting to add an error detail page within gitlab.

E

That gives you details about a specific error and we're also going to add something really quickly, which is unique to get lab, which is the ability to connect the suspect commit, or the first seen commit directly to that error. Detail to really help with error, triage and investigation. But it's kind of a unique thing that gitlab can do by having everything under under one roof and in one single datastore.

E

So those are some of the issues that I'm really excited about in the OP section again, as everyone else has said, check out the videos you can also dive down into the specific links. We love feedback about what we're planning and that's it I'll hand it back to Scott.

A

Here: Kenny: let's: let's move on to David DeSanto for secure and defend David. What are you Thank.

F

You Scout, so I'm David, DeSanto, director of product for secure and defend. First, let's talk about secure for secure, we ate I will say we also have a theme in this release. So I guess that leaves Kenny is the only one without a theme. So far it happens. So our focus is on our stewardship towards the open source community and making sure everybody can contribute, as well as everybody can be secure.

F

So starting with static analysis, as you can see here, we're bringing two of our static analysis scanners down to core the first one is es Lintz, that is a java javascript framework, a second Alice's tool and then nodejs scan, which buys aim is obviously for nodejs. On the dynamic analysis side, the majority items are a carryover from 12.5 I do want to highlight the big one that is coming out in 12.6 and it's the first one here on the board. This is focused on helping.

F

You see your trolls security at the group level on the security dashboard and if we kind of click over here on the big key with this solution, is we're doing collegiate scoring allowing you to see the health by repo and by project, and you can see the scoring right here on the dashboard. It's actually going to be located on the right. If I zoom in on this, you can see that each of the projects have their own expandable. So by default its collapse, and you can see the score for that project.

F

When you expand it, you can start to drill down it, and you can see these 9 projects have addy score so going back. That means they've got at least one high or unknown vulnerability. That needs to be addressed, and then again you could click right on that and drill directly into the project and understand what the issue is. On the composition, analysis side, a lot of really great things coming out as well. The couple I wanted to highlight for you.

F

The first is the Python dependency scanning, so it'll support more text files, not just the requirements, a txt file, that's coming with Python. Also we are getting into supporting license compliance for PHP as composure, doubt lock and then finally bringing dependency scanning to cradle preto projects in Java. So those are a lot of great things that are coming out for secure. If you go to the release page, it's got everybody's been talking about. There's videos there highlighting those items in more detail. For the defense side, there is no major items coming out.

F

Their team is currently working on a long term goals and it kind of show you those you can see those here on our board. The first ones focused on bringing threat, prevention or IPS or IDs to the container, and so that issue there is going to outline that all for you I think you live into that.

F

If you're curious and then the final one is focused on containers, network security and again, the team is working on bringing those out in a later release in this quarter, but their heads down working on these direction items and that's why they show up here on the on the release page as well. So with that I'm going to hand it back to Scott, but you do have any questions feel free to check it out and comment on the issues.

A

Thanks David Josh Lambert will cover enablement and we'll bring us home Josh over to you. Awesome.

B

Thanks Scott, so I'll be talking through our nibblin stage.

B

Our debut on stage is primarily focused on a lot of the operational details of get web, we're a newer stage just taking a second to introduce a series, so we're working on things like installing get lab making mean get lab, making sure get lab has a DRS solution in place in case needed, or a local node for multinational companies to improve performance, multiple areas as well as also ensuring its performance, and that we have our robot API for integrations, and also that we have a great search experience. That's a lot of what we do here.

B

We have a few main things working on this release. In particular, one of them is performance improvements. So one of the topics that we've worked on for a while now was moving away from unicorn, which is a per process model and switching over Houma, which is multi-threaded we're getting very close. At finishing this, up, we've been working to getting puma running on, get live calm for some time now.

B

It's actually running on our Canaria services and we're seeing a little bit of upper flow mysterious right now, and so we're kind of working through that getting more details. But we are seeing a pretty significant benefit in memory reduction. So that's great I'm, just gonna try and do a little more diligence here on this performance.

B

Difference we're seeing between the Puma and nom through my notes, as we continue to improve here and once we have a good handle on that, we're also doing a number of things to lay the groundwork for bringing Puma to all of our customers. It's already available in omnibus as an experimental feature, we're working to identify good defaults, so they can recommend them to our customers and then also making sure it's available and our kubernetes deployment charts as well.

B

So that's a great work there and Puma will also enable some really interesting new features going forward like WebSockets as well, so really excited about getting that across the finish line here. Another key area that we're working on to improve our performance in the future here is moving to you pushes 11 and 12 right now we support 9.6 and 10, but we're pretty far behind the the overall post, guys release cycle and we're really looking forward to taking advantage of some of the new benefits of PG 11 in particularly major improvements to partitioning that's.

B

This will enable a number of performance improvements across the globe instance as a whole and we're trying to drive more quickly to supporting this, and so the goal here is that look at our timeline is to actually have PG 11, as the minimum required push code version and get lab third condado, which is our next major release. Bay. The reason is that we don't want to have multiple code paths or thing get led and by setting PG 11s the minimum. You can take advantage of all these great new features in that release.

B

Please take a look at this issue. If you have some concerns, let us know, but also in the planner you can see, is a release schedule going forward to make sure we stay more date.

B

Here with posters releases as both a and us around using major release cycles, other major improvements, also going on to our API we're working on bringing the ability to have key set face pagination to our API right now we support, what's known as offset based pagination, and so, if you're utilizing this essentially, as you start to make more and more queries deeper into the offsets like page 10, page 100, it starts to get a lot slower and have a bigger impact on your get lab instance, and so we're gonna start offering a key set based pagination, which can offer constant time operation matter, how deep you're going and so improving performance across the board for a lot of our users and we'll be starting to encourage users for deeper offset based pagination by essentially inserting a limit.

B

That's configurable in the instance right around 10,000, as opposed right now to essentially stop using offset based pagination, reduce the impact and switch over to key set best page mention either here in the future, probably around 12.75 any timeframe so stay tuned in there. That's a great improvement, we're also working to improve our coverage, but in support for commit ancestry and listing project services in the API as well.

B

Finally, on the Geo side, we're making a number of improvements here we're making improvements around dr, in particular, with having planned failover testing internally for our instances to make sure we understand the experience and the great experience there as well as we improve it going forward, we're we doing that by also starting to support pausing, replication and maintenance mode here, as well as also the addition of a Geo into our Gillan provisioner, and this is the terraform skits that we use to test internally.

B

Our upgrades so we'll catch any issues that might occur, and they also plan to offer this to all of our customers as official get lab provisioning, terraform scrubs here in the near future, and so everyone take advantage of them as opposed to having to write and maintain their own. So some pretty key improvements here, the lat s1 for geo is they also being just about framework. This is more an internally focused feature, but the idea here is that, as our gitlab feature, development teams go forth and build new features as they do every single release.

B

As you can see here, we need to make sure that we add support and make it easy for them to make sure the data that they're creating is easy to add into the JIRA application and framework, and so right now it's a test.

B

The Geo team takes on at each and every one of these, and so going forward, we'll be adding the support to have a framework that these team can use so that the first time they plan a feature it's already baked in easily with support for geo from the get-go, so some great features there. Please take a look at our videos here, kick-off section for more information, there's a lot more we're doing. That's not covering this kick off and back to you, Scott wrap us up thanks.

A

Off I, don't know about you, but 12/6 looks like it's gonna, be the best release ever I, thanks to all the speakers for doing such a nice job presenting what we have coming, thanks to all the teams behind them that are cranking out value for customers every month, and thanks most of all to all of you who took the time to learn about 12:6 thanks for joining, and we will see you next month. Bye.