OOM Demo

I'm showing that one project kubernetes directory to launched it there's actually only two commands that need to be run one to set up some initial configuration and the one-click script here that is going to launch all of the own app instances just before I get to that though I want to show you the structure that we have here we have each thing that component represented by a directory you'll have project there's a nai spmo and all that's contained in there are kubernetes service and deployment descriptor file your completely with a docker file that are used to create the docker images raid this will look very familiar to you the deployment descriptor is it looks very similar it identifies the daughter image that were trying to avoid environment variable that are passed into the container launches on the volume mounts names themselves and forced that are exposed showing the Morea TV that's used by MSO and this one here the MSO application itself same type of structure with the addition that there's this thing called in ink containers scrub that a little later but it helps us set up identity oh when I kick this off right now and I do one which is the configuration application we do to create multiple instances of own app we give it a name space - I salute it from other instances all the components within that instance under this namespace can communicate with one another annexed to each other when I run this what we're going to see is I'm on a grenade post right now and there is this root directory shared in the cluster and we're going to dump the configuration or the demo in the coming few segments that we have this directory created and as it's creating there's config directories for each of the components and built up here this is the external configuration that has mounted as EML files I was showing you you you I'm going to use a client tool called cute control it allows me to interact with the kubernetes cluster and it tells me all of the containers pods that are running in the environment here we can see that the create config actually ran a container that did the config initialization or the manage space demo and once it's completed as config disappeared the rest of the containers you're seeing are actually just part of the kubernetes you when I was running on here we have the configuration lado so I'm going to go to I can stop here to deploy you it all script and in the same fashion I'm going to specify what instance I'm running what namespace I'm going to run this in so here I have the choice of just deploying every component own up or given this a option here I can actually select which components I want to kick up the death of environment I already bought the whole system I can simply remove and replace one or more of these components what's happening right now is it's walking through those directories you saw and it's watching a few control commands to make the request kubernetes help doing that you look back we have any IFD quality so they're all coming to light now this takes about seven minutes to set up all components and get them into the operational state which you are seeing here is some requests to start the containers are pending and we have these and intestate does back at the MSO that relates to the Annette containers Oh before we actually kick off the MSO container these are like a precondition run a container that waits for Maria DV component be ready this is using two technologies that comes with kubernetes over the box alright has the Annette containers we had we can do pre-processing before launching our main container and readiness probes which are checks to see about whether that component is actually in service whether it can handle requests and once it is unblocked the spawning of the main container look at something a little more complicated like SDC here have actually two and in containers this first one waits for these three back in these three components here yes yes and Cavey make sure that they're ready before launching the next init container which is waiting waiting on the message router on email so the dependencies are results we go on to run to the SEC bagging this now two minutes in we have a number of components that are right now are spilling and while we're doing that I show you use kubernetes but um we like to use Rancher as a way to eat sleep boy and administer kubernetes itself a small cluster that we set up here look at it we have two notes that are running right now look at each node you can save user 32gig systems you can run all of oh not actually on just one blade or VM with 32 gigs everything except DC 80 so that's what's more memory requirements on us I think we require both indicate if we introduced see some of the containers that are starting up here on that stopped which are these readiness probes being Incan diggin it is related to the engine you can see that the CPU is going out memories going up a bit about 13 gigs of the hole you we just recently added a second note to our cluster because we are actively working on the djé linearization and deployment and so we're at six minutes from there or more well T right now as I'll jump in so from rancher I can get two bananas has its own little UI that you can use look at all the namespaces right now and you can see the deployment that a month on ones that are turning up now let the model loader service for any i setting up the other applications are are there you can see that all the components are broken into their individual demo that's what I chose here we can hone in on what is n it's all doing there's number easy to be any application it's running what image they pull down along there's a memory requirements for MSM you okay so we got everything running in a minute minute the canoe and the health check or not oh yeah the Fed is I'll check will fail for dcae because it's not deployed in this area right now on the MS though filled robot test suites Oh some thing is that we've set out or wham up to pull the latest version of the docker images sometimes it becomes unstable in the case of failure because hoping this change what I am doing right now we deployed a portal VNC session air actually that we can use to access the become portal so this is a jump off effectively to get into the own app base so we can access all this opponent intern EECOM portal running it's running on them incriminating certain empty start onboarding oh I think that's about it

Channel: ONAP Project

Views: 969

Rating: 5 out of 5

Keywords: linux foundation, onap, operations, networking, open source

Id: AoomcLgLxbo

Length: 13min 0sec (780 seconds)

Published: Fri Feb 23 2018