 Okay, hi everyone, my name is Mike and today I will show you our tool to how to automate deployment of OpenStack and I'm going to show how we do it. So before that, let me tell you how we did it before. So previously, we were like, first time touching OpenStack, trying to manually deploy it, it takes time and a lot of debugging stuff. Okay, then we wrote scripts, bash scripts to automate that and it worked good. However, it was still a very complex configuration. Then we took Chef and Puppet and wrote models for Puppet and wrote cookbooks for Chef and it is still complicated and moreover, you have to like provision your nodes first of all and then install Chef in it and then run Chef in it. So now we come up with a tool which has a web UI and does all the provisioning of your nodes and then fully automates the deployment of OpenStack and let me just show it. So not to waste the time, I just pre-installed the admin node for the fuel web. The way to install it is just you take an ISO, put on the memory stick, just go to the server, install it in the server, boot the server from it and then you get this on the screen. As you see here, fuel web UI is available at this URL. Then you go here, you see nice UI like this, little notification here that the master node has been installed. What you do next is you just go to your data center and just power on other nodes which must be on the same L2 network and they're going to boot over PXE from the admin node where we got the HTTP server is running and the HTTP and all that PXE kind of stuff. So to automate that with VirtualBox, I've got a script here. Let me just run it. Screen up, this script actually spins up a few VMs which I'm going to use for the OpenStack installation. While it's booting, let me show you what I've got here. So our tool supports a multiple environments of OpenStack. Let's create one of them. And by the way, Grizzly is supported, proof of concept of Grizzly in a simple mode and we got a Folsom release, fully feature supported and stable one. So the Grizzly is not stable yet, however, we are working on it. OK, so first of all, let me browse some tabs while the Discovery Linux image is still booting. So you can choose the deployment mode with support, simple mode when you got only one OpenStack controller and multiple compute nodes and we support HA mode where you are getting production-grade OpenStack environment with HA support supported. So for the type of deployment, you can just use only computes or if you need Cinder, additional block storage, you just go with compute with Cinder. Let's go for a simple mode because this laptop is not that powerful that I cannot spin up a lot of VMs on it. OK, so network settings looks like this. You can specify your network ranges for each network. It supports, right now, Flatshcp and VLAN managers and Quantum support is coming. And we got a nice feature here for network verification when we can send traffic from one node to another, like every node sends traffic to any node. And on the other end, we receive the traffic and that's how we understand the interconnectivity between nodes on each VLAN. By the way, you see little notifications here on the upper right of the screen. It's about the new nodes discovered. You can click on it and get the information about the hardware we got on the node. Then we got a tab where you can modify open stack settings. So some examples here is choose hypervisor type where you can send your logging to some other server in your infrastructure. You can choose scheduler and some other options. So what we have discovered is missing in other tools is to understand what's going on if something goes wrong. So for that, we built a special tab where we can see all the logs of our installation. So for example, you can see the logs from our few web itself. And then let me add some nodes, add them like this. Yeah, let's choose without sender right now. So you can check any other log from remote systems and see what's going on there. Everything is on admin node, so you can get access to it. So for network verification, I can just click verify networks and it will verify it in about a minute. And then in case of something wrong, it shows a table where you can see the node and what the LAN is not coming to the node. OK, let's go and start deployment. When I click on deployment changes and deployment is started, nodes are actually rebooted. And then they're going to load a sentos system from the admin node. So it's going to be installed on the slave nodes. And you see, sentos 6.3 is installing here. So it's going to take a while. Meanwhile, let me show what we can do else. So here in actions tab, we can rename environment or delete environment. Deleting actually means that you can wipe out master boot records from your servers and reboot the server. And then it will load your default discovery system. And these nodes will appear, again, as unallocated. You see here on the upper right screen. So it takes some time to install. And I've got a second laptop where I have everything installed. Let's just switch to another environment so I can show you how it looks like when it's done. OK, so when it is done, you can see nice notification here. And URL, where you can access Horizon. So it is simple as is. You can get to the dashboard. And some other feature I didn't show you yet is configuration of disks. So on a real hardware, you're going to get your multiple disks. You're going to get a multiple physical networks, physical nicks on your server. And you need to configure all that. So for disks, we created this where you can allocate whatever you want for your base operating system, for single volumes, or for virtual machines disks. OK, so the feature for physical nicks configuration is coming. And I hope it's going to be in the nearest months or even less. So here you see two environments, by the way. If you click on Support, you can see here Contact Support. And you will be redirected to our Zendesk system and ask any questions you got so far. And download logs if you want to attach the logs to your Zendesk request. It will simplify the work on digging up the issues you got with this installation. So that's pretty much it. You can install your OpenStack environment in a very fast way, not modifying any configuration files manually, everything in the web UI. And it's as simple as this. By the way, if you look at that environment, which is installing right now, if you go back, so you see here that we have got a nice progress bar showing like how many percentage is already installed and how much is remained. And you can click on logs. And you'll be automatically redirected to Anaconda log if you are provisioning the system or to a puppet log if the deployment OpenStack is in progress. So that's it for now. Any questions so far? You can go to fuel.merantis.com and find it there. What? There we go. Plans for what? Send support? OK. So yeah, right now we support only sender commodity hardware like logical volumes in it. And send support is also coming. It's not going to be hard to create. We need to modify the fuel library to support it and add this into our UI. Can you repeat the question? I don't hear really well here. David, can you repeat that? Yeah? OK, so we got a fuel library, which is a set of puppet models. And we built an orchestration layer and the web UI on top of it. So we got to separate the fuel ISO and we got separately fuel web ISO. So you probably downloaded just fuel. That's my guess is. So if you want to try it out, you can go here. If you don't see fuel web, let me know. I will make it for you, definitely. My name is Mike Sherbakov. I'm Sherbakov at merantis.com. Read what? OK, you can wipe them out. So I mean, yeah, you can delete them. Yeah, where is it? Let me show. You can just go here. Oh, when the installation is done, right now I cannot do this. But when the installation is done, you can click on Delete, choose the node. It's wiped out, reboot it, and slow it with default discovery system. And then you just assign the other role into it. And that's it. It will be rediscovered automatically because it's master boot record is wiped out. And then it tries to PXE boot. And for that node, it will be just a discovery image which will be fitted by Kobler. So it will load the discovery system and again send the information about itself to the admin node. Say it again. Yeah. Yeah, but I think it's for some reasons, so well. So you may power off your node and then, for example, take out some disks or some network cards. And that's why we need rediscovering. And anyway, we have agent running on each node. So it's sending all the time information about the hardware. So if you change anything, we're going to know about it. Why? So, OK, so we need to, for example, we need to create a software rate, for example, configured. And then before we configure the rate, we have to know about the hardware, right? So we have to know it in advance. So yeah, so that's why we need some fast discovering mechanism to configure it and then use it. And again, so right now we can install only on CentOS. And we support only CentOS and few web, few library. We support Ubuntu as well. However, it is like limited version of what few library provides. So in the future, it can be possible to install not only on CentOS, but also on Ubuntu. Any other questions? HA? OK, we got a cluster, Trabit. We got a Galera to do the, yeah, my secret Galera. And what else? HA proxy and keep alive to manage all that kind of stuff.