 Okay, so the next talk is by Jan-Marik Lagoski from the city of Munich, and it's about the city of Munich's Linux system. So please welcome Jan-Marik. Yeah, so hi everybody. Probably you have heard that the city of Munich has moved a lot of its desktop PCs to Linux. That started quite some time ago. If that was quite, it was really quiet for the last few years. At the beginning, we made a lot of presentations where many conferences like this has stopped more or less. So this is the update after quite a few years. So let's see, some city facts just to get you an idea of what's happening in Munich. We have 1.5 million residents since I think last month. There we top 1.5, it's still the third largest city in Germany, 12 large cities in Europe. We have 33,000 employees with PCs and about 1,000 people working in the IT department. It's not just obviously Linux, but we have programming people, we have admins. It's a large community, municipality, so we have a lot of IT people working there. Currently, we have deployed 24,000 PC workstations. Especially if you compare it to the originally 18,000 we started 10 years ago. That's a lot of more machines coming up. Still, that's one of the problems also. We have 50 operation locations in the city, so it's not just a single building where you can deploy everything, obviously, for 33,000 people, but it's all around the city. It's managed by 22 independent IT departments. That means a lot of coordination. That also means normally when the central IT departments where I'm living, that aren't working, that IT at M, when we release something new, 22 IT departments decide on their own when they want to roll out our stuff which we developed. The picture just shows few of them, the largest one. It's still very heterogeneous infrastructure, but it's getting better over time, especially because we moved into a single IT building for the central IT, and that really, hopefully, will help us get more stuff out faster. Just for you to remember, if you have heard of a project, the project has actually started in 2001. That was, at this time, an internal project. The IT has realized, okay, we are on Nt4. Microsoft's saying we don't support Nt4 anymore, and then IT asked the IT council, what should we do if we cannot use Nt4 anymore? And the central city council then decided, okay, what options do we say have? And then there was about two years catching up with ideas. Different companies came, different ideas came up, and at the end, it was decided, okay. Independence is good, better than depending on some larger company, and so let's try Linux. The original project goal was to migrate 80% of all the desktop PCs, and just remember that that's the desktop. We are not talking any servers. We were not consolidating anything. The heterogeneous infrastructure was still there, and the city decided to roll out Linux clients, so that's what we were facing at that time. That means we had some Nouvelle servers, some Solaris servers, or whatever hardware was built in the last 20 years was still standing somewhere, and we had to support that from the Linux client side. Then 2006, the first beta was released after a long tender, and while we are already developing Linux, we realized that we can also start migrating people to open office on Windows before actually they get a Linux client, so it's much easier for them if they already know the programs that are working them. Most people don't really realize that they have Linux and they do not really care. That was something important we found out during this time, so even when some people were complaining about Linux, you were looking at the PCs, and there was actually Windows XP running on it, so really people don't care. They need their office, they need their Firefox and their Thunderbird for web email and do their stuff, and they want to do their stuff, they don't, normal people don't care really. So in 2011, half of the expected PCs were migrated, 2013, but Linux project was really finished, so internally it's just supporting Linux in the city. Linux project, usually, officially are not really allowed to talk about Linux anymore because Linux is finished. But Linux is still growing in city-wide use, so we are currently having 18,000 PCs, and if you look at 2005, there was a maximum of all PCs was 15,000, so we are more than the maximum. That was there a few years before. That's the current state. So what are we currently deploying? Current release is our 5.0 release. We are using Duck, which you might know already. We started using a little, we were really little with ReprePro. It works if you are just single or two developers, but if you have a few people uploading stuff, it gets confusing and there's not very good locking in ReprePro for databases, so we implemented our own version of Duck with a few additional features, which will be in upstream Duck, hopefully at the end of the year because we get the student and he's going to port all of the stuff to Deviant's official version, so hopefully that will work out. Yeah, we have rolling out Firefox and Thunderbird ESR. We are rolling out updates for this ESR version, but we have to stay with ESR, otherwise a lot of applications may break which the people are using. And our most patched program is definitely LibreOffice. We have on 4.1, that's already two years old, and we have something like 300 patches on that. So that's really more or less a huge patch rate, but everything is in master, so everything we have developed is already in 5.0. We just did back port after developing this stuff. We make Q&A after the talk, please. So yeah, at the end we have 296 changed source packages, which have changes from us with own patches, back port and everything, and a total install base of about 4,000 binary packages. So now you have 18,000 PCs, what do you do to manage that? That's something really, at the beginning, it was really hard. There's those LDAP management tool called GOSA. It has a lot of features, but it was not originally designed to manage clients. That was something we were mainly developing in 2005 and 2006. It's open source, you can get the current version at GitHub in this repository. I can, if you are really interested, I can show you that after the talk, but not here. We install the stuff using FAI, so it's completely automated. You can trigger the installation from GOSA, you can put a date when the computer should be waking up, and basically all our installed data is in LDAP. There's just one single reason for that, because LDAP is the only city-wide usable database. Everything else is so independent that you can't use that. Client and user management infrastructure is also developed by us. We generate application preconfigurations, unlock in KDE menu, host settings, FAI integration. We mount network shirts with user credentials. We manage the printers through the OS file, so admin stuff. We have a roaming profile, so you can log on on every PC in the system, but it's static setting, so you don't have a home directory on a share, but it's already, also it's synced. We have exoc preconfiguration, and we manage USB storage device, so users cannot bring their stuff in and copy whatever they want. It's Python 2-based to our current release, and it caches the stuff locally, so even if network is down, people can still work, log in. And yeah, it's not yet free, so I hope a lot of people will write emails so to give back interest, and then the city will release that stuff. We did that stuff with Walmux, but that's really the only project we released. So we have just an idea of what we are releasing. We have four main targets, Buzzer's client with the normal PC for the user. It's fully managed, it gets all the stuff to install, it's completely managed by Goza. Then we have notebook installations, that's just basically the normal client plus a few stuff which are just used by notebooks. These notebooks can just be used in the internet or offline, otherwise they are not usable. Then we have some, what we call, Outback client for presentations, they are not allowed to use in the city, but you can copy stuff there like for presentations here. And then we have a self-replicating install server, so we have the infrastructure distributed to the 22 departments, and we managed to distribute the load for installations, because as you may know, our file uses NFS routes, and we use NFS4 since the beginning, it's already also now included in FII, and so we need local installations, almost local. So just because we're in the Wikipedia, I think we lost photo of the client who was from 2006 or something like that, and every time that people looked at that, so you can see it's 2015, eight, three. It's just a normal KDE, nothing special. I just put the menu there to see anything. There is no magic involved, that's how people are using that stuff. So, and just to give an idea how long we are working on that release, we are still on Ubuntu, obviously, but we started on 12.04 in 2012. We had the first point release because there were so many bugs, okay. So took just one month, about three months to get to the software distribution initial server to update, then we started our normal development circles, then we realized, okay, we have to put a lot of more time to get LibreOffice into a usable way for our users, then development finished, testing finished, and the final release was almost 2015. So it's a really, really long development cycle, and we want to cut that down, but we don't currently see much opinions because all the time when people are added to our Linux clients come up, we have more support, we have more idea what to implement, and so that's currently our development cycle. I hope we will get faster with the next release. So what's our really major problem because Linux normally works quite good, but as you have seen, our development cycle is large, our release cycle is also large. So for 80,000 PCs, we stock new hardware for about half a year, that's still a lot because for 80,000 PCs, you can think how much hardware breaks per day almost, and after half a year, the hardware stack which is delivered with a boot release or any other Linux release is normally too old for new hardware. What does that mean? It means you have a back ported kernel, you have a back ported DRM, you have a back ported Mesa, Xorg, Xorg drivers. At least you get that for one half of a year, for two years from Ubuntu in this way, so that was one major decision why we originally started with Deepian and then switched to Ubuntu because we couldn't come up, get up with the hardware and release cycles were just too short actually for us. We would like to even have longer release cycles but two years seems to be okay, but our own development took last year for the last release two years too, so it's kind of critical. The second thing is even if you use a hardware enablement stack from Ubuntu, you have to use one which is supplied with a distribution or one with two years later. And every time you have a new version of your hardware support, you have to check it on all the hardware outside. That's really a lot of work. And a lot of work is not done in the central IT but in the distributed IT because they are independent. And normally sometimes it happens quite often that they have stuff which doesn't show up with us and we just learned that after we roll out actually it has hit and it doesn't work. So we hope to cut that a little bit down with hardware enablement stacks but just a new idea which came up. So for one last slide, it's just what will be happening in the next year. It's their currently IT projects in the city of Minidap and probably we've heard of migration of mail and calendar clients that was also publicly available but not so common now. We'll be the first one which have a large contact windows rollout. We hope we can get this upgrade for 4.0 from 12 to four to 14 to four. And then really have a new one, new release based on KDE5 system B64 bit but that will really get a lot of work. We get a new Gosa version. There's also the link and we go to new infrastructure which will do all the stuff for us. So yeah, further information, email. There are a few English pages but I think email is still there. So if you really want to know something new or have some ideas, just contact us. There are sometimes people visiting from other countries. So we are open to that. So if you have an idea, want to roll something else, want to get some more impressions, write an email, say, okay, we will come in North Dakota, probably not in the next week but in the next month and we can organize some views of that. And yeah, that's it. So thank you for your attention. Thank you, Mark. Is there any questions? Hello, you had quite a lot of patches for LibreOffice. Did you work with upstream trying to merge them or submit to the LibreOffice back tracker? And is it that in your future plans also? Or somehow can we help you out to do that? So we are actually involved in the, I'm a LibreOffice developer. I'm a responsible KDE4 backend. All of the bugs and patches we have in current LibreOffice, everything is fixed. So we developed those 300 patches most of them and backported the stuff for our release because we could not switch. We just decided at some point we stay with 4.1 and get that ready. Otherwise you get a lot of more bugs again. So everything is there. We are developing with the community. We have support. That's fine. Other questions? How do you handle synchronization of data for your laptop clients? How do I synchronize that? For example, use of home directories and the configurations that they've got. If they are on a laptop and not necessarily on the network, then how do you deal with synchronization of, I guess they have a central storage somewhere and they have what's on the laptop? Normally you have normal network shares and then we develop known synchronization process which means on lock-in your shares mounted. Then from a loopback partition, because it's a sift share, it's a Windows share, you have a loopback partition there and then you have an air sink running to get all the stuff to your computer or Windows. It's independent if you have a notebook or a local PC. So you can practically go everywhere in the city, sit down, wait something like five minutes and then all your stuff is there. You haven't mentioned any kind of group where actually, but I saw the Oracle calendar. I mean, is that the only thing that you have built in in terms of the stack? So the question was if we have something else currently or yeah, that's a current project which I just mentioned, it's my mic mark that also was a tender like three years ago or something like that. We are moving to Colab completely and this will be finished 2016 somewhere at the beginning. Okay, I think we have time for one more question. Hello, for the authentication and stuff, are you using this or are you managing to do it with just Kerberos and Dell-Dap? What? Yeah, it's... No, everything is an add up. Even without everything really, it's add up based because that's really the only central database we have. So it's nothing special like that. How do you cache the credentials for the laptops? So you just with a laptop, you lock in and then your credentials are cached locally. Everything is basically with our client management infrastructure, you can lock in, then your data is encrypted with your login password and then you can use that everywhere. It's just temporary cache locally. Thank you. Also, we're running out of time but Jan Marek I think will be here for a couple of days and some other people from the city so just catch them if you have further questions. Otherwise, we'll be on with the check committee panel in a few minutes. Thanks a lot again. Yeah. Yeah.