 Welcome everyone to this month's edition of the platform backend functional group of date My name is our man, and I'm the engineering manager for this team and today is the 30th of May 2018 I'd like to start this edition off by welcoming Imre Farkas, who is the latest addition to our team. Hello. Yes, there you go Imre Farkas is the latest addition to our team. He joined three weeks ago to the day Imre, would you like to take a moment to introduce yourself to those of us who might not know you yet? Sure, thanks for the introduction So my name is Imre Farkas and I'm located in in Budapest, Hungary I previously worked on OpenStack on the permittal management service and also supporting operations Later I joined the company providing a marketing platform with AI and big data solutions and then three weeks ago I joined GitLab and Previously I've used GitLab on my private server and hosting my repositories and I was quite happy that I have a such a great tool that just worked and I was even more happy to learn that GitLab is hiring So I started to learn more about the company and I got quite enthusiastic about the OpenCatcher and of course the OpenSource product, which I have used for long So so far I'm still getting up to speed and Working on smaller issues and on general on-boarding which I have enjoyed a lot so far So that's it for me All right. Thank you, Imre. I'm very happy you found out we were hiring because I'm very happy you've been able to Join the team All right next up Let me see since we last spoke which is about five weeks ago in the last functional group of data for this team We have done a couple things I'm not going to go into too much detail for the first three But I like to be complete for those of our viewers who might not know that a GitLab development cycle as well as most of the team members do But one notable thing is that on May 22nd We GitLab released GitLab 10.8 which had some notable contributions from the platform team First of all we moved push mirroring to core which means it is now open source and available for free to everyone And the great thing about this is that it will allow people to use their GitLab repository as a canonical source and then still have a copy Read-only copy on another instance because for example, they already have an existing community there or because they're moving away from it They can do that in you know kind of a steps progress and process instead of doing it all at once Next up for GDPR we had to scramble a little bit But you managed to get the functionality inside GitLab to enforce acceptance of terms of service We managed to get it in before the 25th when the GDPR went into effect Thanks Bob for working on that We also improved the performance of the repository size limit check GitLab has functionality where you can limit the maximum size of an individual repository in GitLab comm We wanted to enable this to limit repositories to a maximum size of 10 gigabytes But the actual check that Performed when a new push came in that verified whether the incoming push was larger or smaller than this size limit Was relatively slow We which meant we never actually enabled it because it would just slow down pushes so much with this improved performance We are able gonna be able to do that if we haven't already. Thank you Ruben for working on that And then last but not least the title might not sound too interesting when I tried to explain what the point of this was In the previous FGU, I don't think I did a very good job So I hope that the next graph will be more telling Can you spot when that change went live that change to import to move import and mirroring related columns of the project stable? I think you can figure it out and if you look at that about May 10th at 4 p.m You'll see that total debt tuples went from about 20k to somewhere around 5k For those of you who are not database or specifically Postgres experts Basically debt tuples are bloat inside the Postgres database When you delete a row or when you update a row these rows are not immediately deleted from file system They are simply marked as debt Rows that shouldn't be looked at anymore and they will be around until next next vacuum goes around Vacuums can be relatively costly and vacuums can block a lot of other stuff that's going on So you want to do a few little of this is possible And you don't want it to take too long and on top of that the more debt tuples are inside the repository I mean database The larger the actual storage sites on this will be and the slower any queries will be because of course They will be faster if they have a smaller data set to look through than if the data set is larger And these import and mirroring related columns were a huge part of the you know kind of reason why we had so many debt tuples in the project stable and Thanks to the Great Work Tiago did here. This is you know Now only a fourth what we were seeing before after this change went live So that is very good for get left with calm and very good for all of our experiences using it So that's what we've been working on the last month or so Until the next time I have an FGU like this. These are some of the things we'd like we're gonna do First of all, we're gonna finalize development of get lab 11.0 on June 7th and release in 20 seconds There's a couple of things that the platform backend theme is contributing to this First of all, we are going into general availability with group level single sign-on on get that calm using Semmel This is an effort that has been in development for a few months already And it's been live behind a feature flag for a while as well We've been kind of testing it and now for 11.0 It's finally going to be ready to be opened up to the world This will be very interesting to those larger customers or companies that want to use Semmel to manage their Provisioning and their their users and their memberships But who do not want to actually host their own get lab instance now They'll be able to hook up to Semmel to the get lab.com group and they can use single sign-on there More of us error handling and we're trying of fork and import failures Of course, we don't like it when forks are imports fail So this will definitely help there and there's a couple other of major efforts that we are contributing to 11.0 Also notable to anyone watching will be that we will be removing API v3 We deprecated this in a in give up 10.0 when we launched API v4 We were originally planning to remove it a couple of months after that which would have been like six months ago from now, but because of course for larger Customers and environments that can't take quite a while to modify all of their tooling and all of integrations We have kind of kept postponing this and then at some point we decided to just do it in the next major release Which is happening on June 22nd So in get lab 11.0 API v3 will be completely gone We'll also be deprecating DSA SSH keys because these are Relatively unsafe and they have been deprecated inside open SSL for a while Open SSH I mean and now we are also going to be deprecating them on the get lab side of things which means that we will Prevent people from creating new that DSS DS a SSH keys And we will probably still allow them to be used for a transitioning period people already have their DSA SSH key set up, but at some point we will also stop accepting them entirely and of course we'll let everyone know when that happens We are doing a little bit more in depth Of course if you pull out that more link you can see in our issue tracker exactly what the platform theme is working on for 11.0 and Then on June 8th, we will kick off the development of get lab 11.1 Oh, actually Remy says v3 was actually deprecated starting with 9.0 not 10.0 has it been that long really more than a year and a half That's insane. My memory is It's not so good anymore. I guess Either way, I was saying June 8th will kick off development of get lab 11.1 and what's gonna go in there We will only be determining ahead of that June 8th and you can follow that life with us in the kickoff on that 8th of June That's it for today's function group updates. Are there any questions from those in attendance? Looks like there aren't so everyone have a great rest of your day and you get 22 minutes of that day left back Cheers everyone. Thanks for your attendance