Jenkins X Office Hours, 17 May 2018

Previous Meeting

⏯

youtube image

►

From YouTube: Jenkins X Office Hours (2018-05-17)

Description

No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).

A

All right welcome.

B

This is our our first Jenkins X community office hours awesome and we have a key. It's not just three of us there's four of us.

B

We have salable there as well. Yes, we've got one more person, brilliant.

C

So we're looking for opinion barthelona.

B

Yeah we're getting common set, okay cool. So what do they do? In this slack channel I'm just going to drop the Google Doc we've got so we're just going to try and follow some quick people have some questions or anything you want to add in to their requests for future demos as well, and it's really just to track some information share some ideas as well, so for previous meetings as well.

B

So, oh hello, we've got something else. There's our so I think we're going to try and maybe kick off just by say, hello welcome. This is this. Is the first meeting we're having and hope we're going to do this for Evie every two weeks, every Thursday providing it works? Well, we get people, it provides benefit for people, I think the aim is to try and create to start coach, creating an environment for people that can share some ideas. Some some experiences- it's not just from ourselves, but from people they're actually from users as well.

B

People involved in the community or people wanting to get involved in the community as well. So we've got some ideas so gave this because he gave in oh there we go I can see. Let's see you know all right there, so yeah I mean I. Guess we could start them. We mean James, have both been working on that some. What we think is some cool new features this week.

B

It's this week's new flavor of fun week, but we can probably save that, if maybe for a little bit, if you just may want to open it up to see if anybody's got any questions just to kick off with I guess.

D

C

D

No questions here just trying the way the web tools for spring. Oh yeah, awesome yeah, give it a try to that, trying to understand a little bit more about that. I haven't used it before so trying to catch up, believe it or not. Yeah.

E

I always forget how to do it. I have one project and then completely forget and the next hour I'm like what did I do again? It's like a there's, a dependency. You add and I think there's some other things. You add and yeah I mean it's really cool. When he works. I always forget the steps you have to go through to make a project work with the dev tools, yeah.

D

I'm still thinking all the sink that sink, it's going to be tricky so on yeah, yeah.

E

So that so the way we're using death pods is when we're doing things the long way we're making a brand new docker image, and we just really took the new dock image like we'd, run any other image with the spring dev tools. The aim is to run a process and then that watches for the jar you know reloads the jar on the fly, so we don't have to make a new docker image, so we can just kind of copy the new jar.

E

If you rebuild your spring boom jar, you can't just copy that from your laptop into a running pod and.

C

E

You have the right, like environment variables and whatever the command line flags or whatever these big spring boo acts in this watch mode, and it will watch the gern if the jet change, we load them a fly. We.

D

E

Pushing the docker image and stuff yeah.

D

But okay, but the dev tools needs to be running inside the pod right and in my.

F

D

Same time, yeah, that's fine yeah.

D

E

You I think we probably want to have like a maven profile to add the dev tools jar into the spring boot to you to build it in dev tools mode. Then you've got a socket image with the spring burlap inside that the dev tools are running in there. Then we need a sink command, that synchronizes the jar into the running pods, which is a little bit different than the where deaf pods work, but we should be able to use the same casing to land the covers, so that could give us a really nice fast rapid.

E

You know type some curd cereal, but on the fly with spring proof spring GU dev tools has got the by the live reload stuff as well, so your edit, if you're doing a web app and you edit the CSS on the HTML, it will reload the browser on the fly. So this dev tools are really cool. We just need to figure out exactly how we make spring move. Dev tools, work inside Jenkins X, but it shouldn't live to her.

E

The hardest bit is conditionally, adding the dev tools in so your palm des XML means I can profile or something to have the dev tools. Yeah.

D

Yeah I'm going that direction. I will keep you posted about my findings, yeah.

E

Awesome awesome, but I would love that I think we're gonna have to so the dev pods all working kind of the same way, because it's language, agnostic but I think when, when it comes to spring boot we probably won't have a special way of doing things where we don't keep rebuilding docker images. We just run the pod once and then we just copy the jar into it, just to get that more rapid, cheaper I mean even even when you're doing that it's still not lightning. It's not fast. Yes, it's big fast would be.

E

The word would use it's, especially if using things like hibernators can take. You know 30 seconds just to start up, so it can be like a minute before the app reloads. If you buy J rebel, then that can go really fast, because I can do incremental class loading and stuff. So if you only you know, change one class, it can just reload that class. So that can get better.

E

So we should maybe add in there and add on so that if you bought your Apple, we can include J ripple in the running thing, but we could probably do that in the profile, don't kind of a vag, a rebel profile that we just add a rebel in that's well, but I.

E

Ideally one of the problems with maven we've got Tommy to jar from maven stuffy already, but one of the problems with maven is there's no way to use bombs for plugins mm-hmm. So it's kind of hard to say generically. Add the spring dev tools capability on any project. You have to kind of modify the pom.xml, but we could we could get Jake's import to do that. We could say.

E

Oh, this is a spring boo, app, let's add a profile, but doing this, the bookie kind of stuff, which would be kind of nice, yes, Abbey's spring boo is one of the most popular Mikey. So these feelings around so it'd be nice to have a really awesome. Out-Of-The-Box pingu experience really yeah.

D

Totally another thing that I might say around that that space is that I've been testing the new spring called gateway with the kubernetes service discovery, all right, so without deploying you riggidy in there, so basically just lifting in the currents, registry and I got it working. So I will share that examples. Well, folks, people that might mean.

E

Yeah, if we can use static DNS, we talked about this a little bit on slack today. If you can just hard code all your service names, so you talked to cheese in the air or whatever nuts, that's just it, and there's no dollar squeaking is there's no Cuban, 80s API or anything life's, really really simple. You can write services that talk to the keeping a REST API, but I think we found this one slightly something to do that. You then have to add the library to talk to cuban. It is then yeah.

E

That's right at the service account. Then you need to add the roles in the service account so that you can query the various resources in keeping the user. It can take a little while to tweak your application, so it can be.

D

Yeah, but imagine that this use case will worry where you have this gateway, that automatically register routes right, so that's kind of a place where you will need that so I'm just trying to figure out. Where are the tweaks and it's working? So it's you know it's even more complicated when it works. Yeah.

E

I mean it'd, be nice to have a generic app. That does the route so that, if your micro service needs a--, you don't have to mess with all this service account on roads and stuff. So you just have a generic router thing. Either takes care of this for you in many ways, this deal is trying to do that as well, where in history you just talk to, you know canonical points, logical, DNS names, and then it does all the magic behind curtain to figure out what those really map to whether they really are that service.

E

All you know it does. Customs are bouncing in the front, yet yeah I think one of the things you all kind of want is it's for. Cuba net is to make things easier whenever there's complexity like custom load, balancers or whatever, we can wrap that in the micro service and hide it from the app so that the app everything from the app looks really simple.

E

They spoke with two simple endpoints and if you have a clever little balance over them and that's a separate micro service- and you hide that complexity in another place and hopefully those load balancers can become generic. Because most of those are, you know, just clearing services and looking for label selectors or whatever to just decide where to go. But.

B

The apps there are to talk in a consistent way the same through each applications. The micro services, then that actually handle the cool, exterior yeah.

E

One of the things I really like about the way Cuban is tries to help us make basically dances.

E

There's if we assume DNS is a thing and that could just talk to you, know, cheese and beer and wine or whatever the services are called, and that could just make up names of endpoints it wants to talk to, and then you can do kind of dependency injection to map those services into something real.

E

It could be just a DNS name, it could be a pod selector, it could be an external name or it could be a micro service that there's a routing for you, just like your example with a gateway, but from the ass perspective, it's just a DNS name. You talk to him, you can't hide their complexity suits. It feels a bit like dependency injection, but but at the network level, rather than at that spring bean level.

B

Cool, so um so it's just all Gareth's joined this down. There go some wave guys there. You go so guys join us in there. In a couple of weeks, Owens there as well, who were I, met in Cardiff he's you guys did the the angular QuickStart so which is in them and I, think they're working.

B

So thanks very much for that, so we could put some things we can show. But does anybody have any questions about what' anything that's going on at the moment, I saw some Jason pop-up there and I think he's having some issues at the moment with some nexus settings up, Oh Jason's back, so don't if you can hear but we'll have it. Maybe you have a look at that on skype after the call, but if there's any but any other questions anybody has at the moment anything that's ongoing anything they want to share.

G

Yeah I've got a question, awesome sure um so I'm wondering if it's, if, if it's possible, to have an external chart registry or how hard it is to get that running, I haven't had much time to play with this in the past couple weeks, but um I'm hoping to get started again next week or two but yeah having an external registry, and then we could restore an environment from you know with without having to rebuild everything in the cluster. So.

D

I'm using an external registry, basically I'm registering one in my make file right.

E

Yeah we do the same ourselves, which you can just do the you know, helm, helm, repo, add URL in your make file to make sure that the build pod knows about the remote registry and the one tricky bit is, if you need secrets to access it, although to be fair, I think yeah. If it's an internal URL you're fine, but if it's public you might want to protect your own. So we've probably secrets for those.

E

B

For the church, art museum register registry repository yeah, because one of the things we've actually, we were actually using persistent volumes and one of the things on the to-do list is actually switch. That out, when you are say, a KS e, KS or g ke actually use that the native cloud storage rather than persistent volumes. So then you can actually totally it gives you that kind of accidental registry itself. You can totally destroy your your cluster, but then you can recover all of your previous charts.

B

For example, when you restore it, I think that one of the other things it's on the to-do list, I think chart Museum, certainly supports that. It's just we haven't configured that for Junkers exile to the box, yet.

G

Okay, nice and things.

B

E

So it's going to link.

B

E

From Jacob a how hard would it be to integrate chair, it should be reasonably easy. There's a couple of issues already about integrating cloud Lionel chair, so we added death pods, which are kind of the first step towards ID ease, and so we have build pods, which are basically that pod templates used by Jenkins to do conc do pipelines. So that's where you find all the tools you need in your pipeline, so whatever version of mayor Furman, cube, CTL and power and get and scaffold and all those kind of things you need to do.

E

Cnc deep and we've always kind of wanted is to align the tools that developers use and see ICD uses so that we're always using the same. You know same maven, the same hound, the same cube GTL and this you don't really want everybody's laptop to be managed separately, but then the cluster having a complete different version of everything. So we kind of want to align those. So the first step along the way is dev pods, which outputs the link to the chat for depth pods, which is basically just a way of reusing.

E

The same pod templates for doing developments you can literally just create a dev pod and at that part, is just a pod for you to you. This developer, to you run, may have and run node do builds or whatever, and you can do incremental builds through scaffold.

E

Now, I did a little spike yesterday, if the before yesterday of, could we have a command line? Option that when you created Airport, could you add in a sidecar for an IDE and I did a little prototype with cloud 9, which kind of nearly worked?

E

It was there almost usable, and so what we're thinking is we could have a dev pod of and have pluggable sidecars for in a cloud 9 at and maybe one day, vs code Eclipse chair, whatever IDE, which uses a sidecar, then the sidecar for the IDE and the deaf pod could both use the same shared volume for the stuffy, editing and I think also just kind of worked. The only thing we need from in the Christian perspective is what we'd really like is.

E

The deaf pod container has all of the tools that we want to use right. The maven and scaffold and all those key things we just want to get Eclipse chair to use the same container for running, builds and stuffing. So we want kind of one container that runs the web IDE bit and then all the commands we want those to run in the other container if we can get that to it with a clicks chair.

E

That would be awesome, and hopefully we can do the same with other editors as well, because people like all kinds of different ones, but yes it it shouldn't be too hard. I, don't think so. Hopefully, someone can help.

B

Mcnutt serpents, so with that, would that mean because one of the things with a fiasco stuff that we were looking at really wanted to be able to make a pull request? You know with our preview environments, then have a link to the the editor. Really that's. What I want is the IDE a link browser base with those changes, kind of running somewhere, even mountings, know.

F

B

Some test repos, for example, that you can kind of. Maybe then, when somewhat simulates tests over yeah, that.

E

Will be all I desperately to be able to just edit a pull request and just open the browser and you're in the pony quest was all the code there and you don't have to do all that bit magic to cannons the branch fork Paul. That would be just awesome because you often just want to tweak a pull request right. That's for me! That's the that's the killer use case for my based IDE I.

E

Think a lot of developers are quite attached to their heavy-duty IDE right like a you, could enough to really pull IntelliJ away from me with I'm gonna be holding on with my fingernails with the last breath, but but there's times when I use, you know that github UI to do some changes because he's just a simple change and it's easier just to do it in the UI in it, so I think for pull requests, particularly when you know something goes wrong with the pull request, just hitting a pull request.

E

That's a perfect use case for web based IDE, and that's just a question of which one is it easy chair? Is it? Is it cloud9? Is it something in the item vs code ecosystem I saw another one we over there that let's come along, it's just great of which one is it going to be bird? Yes, I would love a web-based idea, progressed just one other thing about death pods.

E

So right now, whenever you push a pull request, you get a preview environment. What we kind of want to do with death pods was have a way of deploying things on a per developer basis before you've done a pull request.

E

So what we kind of do have I'm, not sure about the name of this. We call it Eddy environments, so each user gets an edit environment. We should never call it a personal environment, very newsroom or something but basic. Each person gets their own environment that if ever you run a def pod commands and you run a scaffold. Build scaffold will then deploy automatically into your personal namespace. So if two of us have got dead pods at the same time, we both do a build.

E

Each of our builds will go in different namespaces to me that kind of conflict. It would be nice if we got say something like chair working, we kind of want chair on cloud 9 and with obvious code. We kind of want those to use the same underlying commands to do the build, so you know scaffold their scaffold run or whatever the commands end up being to do the build and run, and if you get that working in an IDE, I think that'll be really really awesome.

E

I mean today, like I've done demos recently with death pots and I've used a desktop IDE like vs code and use JX sync to synchronize my desktop editor with the death pod. If you in a web-based IDE that could just be in the death pod and then we don't need to worry about synchronizing, you know they can just be one persistable um-- with your changes on and you can edit it in the death pot either with vy. If you want to keep it out code or with a web-based ID.

D

Awesome can I ask one question that might be related to that and I'm trying to figure out. It might be a very silly one, but it's related with something that also I. Don't know. Change also asked industry about things right. So how do I make my team to connect to the existing cluster around the documentation about that right? So, instead of doing instead of pushing them to create a fork of the project and create their own, you know pipelines for that how they can join my pipeline. So how do we do that I mean?

D

What's the recommended approach for that, or maybe you know if you are planning to do something around it, it would be interesting to know what's the direction yeah.

E

We would need to polish that experience a bit really I mean today. Jx is using the same Cube config. To talk to the cluster that cube. Ctl is so that I mean the first step is connect to the concept to the cluster, just like you do in say, Google, Google, Chrome or whatever, and then assuming you have access to the new space that has the team inside. You should be able to just use all the JX commands.

E

We should, though, how we could do even better commands to do. For example, it would be nice to be able to say clone all the repos for my team into the file system, so they're all just there, so I could just do it rather than finding each get URL by hand just kind of going clone, my repo, so all the team's repos a clone.

E

You know in a folder, so you almost have like a workspace that gets set up for you in the file system with all of the pipe all of the projects, maybe even focus them all automatically server you're working on a fork.

E

Some people don't like to work in the foreground can make that configurable, so it would be nice and stuff a single CLI that just says you know, join team or something team sets up or something, and then that queries, what are all the apps or what are all the apps in your team, clone all the repos to the file system and then maybe even open your ID you'll know more. You can manually open them all individually or whatever.

E

That would be really I mean if it's super simple to do because they're all in CR these we know we know what they all are, so that'll be really really nice. Yes, I'd like to do that. If ever we go web, ID I think it'd be really easy to be able just open editors directly from you, JX or from the biggest console or something I think that'd be really nice. Now let me make a note of the club called Reapers. That's a really nice idea. I've.

B

Been popping some notes in the G key dock and more thank yous, although I didn't take that one now, there was a question that you from knowing about well. Is there any kind of is a way to have an audit tool, so they can look for bad practices of clusters, so the example was giving things of like encryption and rest for non secure, API server, for example, I. Don't know in that particular example. But the things I spent in my mind was sauna boy, but I don't know.

B

Is that what you would say, James yeah I was gonna.

E

Give the same answer so one of the problems we have in general with well. Everyone has in general with anything that's keeping its best. Is nobody quite knows what a cluster looks like, and so we build all these tools and we kind of assume there's a nice working posture, and we just write this. You know yeah molar ham, charts and we assume everything's gonna work, but then there's all kinds of different clusters out there, and sometimes this well behaving keeping it exact, isn't work on the cluster.

E

So having a way of just verifying is the cluster set? Up correctly, you know its DNS working and so forth. Yeah we've looked at so no bones, so nobody looks a really nice way of basically just running a bunch of compatibility tests to go. Yes, DNS is working. Yes, we can talk to if I am in the ingress and all those kind of things I mean it's so easy to mess up keeping it is really. Oh, not myself have a Cuban. It is class, so that's configured so that it won't work with Jenkins X.

E

What we kind of like to do is have have a very declarative description of exactly what Jenkins X needs like we need role based, authentication enabled we need an ingress. We need to be able to talk over dns into the cluster from outside the cluster. To you know, work with web consoles and stuff. Some people want HTTP inserts having and murmuring, which is great, although people just want something that kind of works to kick the tires, so it'd be really nice.

E

If we had a way of describing what the user is trying to do like, is it HTTPS or not, which DNS name whatever, and then we could run so avoid to validate? Is the clothes to behave incorrectly of the various things that like Owen said it is the API server secure, are the search for everything and these everything HTTP only and all that kind stuff.

E

Another thing that checks Philippe's today we're using insecure doctor registries, which might scare people and think, oh, my god, things are insecure by default. All that literally means is. We can use a docker registry in the cluster to store images and we only access it by a service IP. So it's not meant to be on the internet necessarily or doesn't have to believe that if he was on the internet, then it should be secure over HTTPS, but then there's a problem. It's kind of hard to do a talk to a local registry.

E

That's running inside keeping it easy on the server's IP, that's not public, so there's no public certs. So it's kind of hard to do that securely. So we kind of today the out of the box, is to use an insecure registry. I. Think one thing we really need to do, though, is every time I show. Jake is extra people, they say. Oh, this is how I make my clusters and everybody's got their own shell, scripts or ways on terraform or whatever it is that they used to make clusters.

E

We kind of want to bed drink execs insulation into whatever people are using so say terraform or a shell script or whatever, so that we can have like canonical ways of is that you make you install jinkies. Ethically Cox on Amazon Paul G cloud with Google Hawaa, whatever so that we don't have to mimic what people are already doing in Jenkins X they can keep whatever it is. They do to make posters so Google folks always use g-cloud and it clusters cops.

E

Folks, always use cops, keep the cops and the G cloud native and then just layer, the the Jenkins X install on top, because we should be doing guitars to manage Jenkins X, which is one of our kind of things. We cut a bit of a corner there just to get something out, started I'd really like us to have a get ups way of managing and installing Jenkins X. So.

B

Just just throw this so that's the actual, the initial installation there we're actually putting through an ocular nexus Jenkins, for example, and if we add an add-on. Currently we just kind of apply it to that by development team. What we're talking what James is saying yeah we need to have that all backed by get as well. In the same way, we do our other environments, but.

E

This one is coming, it'd, be nice to put teams in there and users so that we can.

F

E

The five users on my tea I've got three teams: ten people each and that should be able to set up, but all the our back for each user in Cuban is that should be a setup all the organization since a github bitbucket or whatever you're, using give each user the reading right access to the repos in all the organizations and so on, so that and if ever you want to add or remove users and teams, you just edit the get people and get ups and get terraform, for example, to reapply your your settings, so I'm really keen to get a nice terraform installer where everything's declarative and you can use whichever terraform add-ons you wish to use from making the given use cluster and and everything else just becomes kind of automated just to bootstrap the cluster.

E

This is like once the clusters running use Jenkins exiting entity for all the apps, but this is more just to spin up that. It's one of the initial teams, I think they'll be really nice and.

B

Because some some help with that at the moment haven't we so, but that's actually kicked off a little bit. I mean there's folks from the community that have come up with terraform expertise and I. Think we got in March sometime next week as well to try and iron that out as well, so hopefully that they will get missing.

C

As well, so it's at least three forward to validate deployment for you, so I mean, which would be handy. Like he's, performers are opinionated way of like in sorting thinking, decks, yeah.

B

Yeah yeah I popped a link to sauna boy, that's from the head folks from here as well. That's in the chat and they've also got some other really cool projects as well, so I recommend people check them out as well. Obviously, the one gimble I always forget that.

E

That's the look, I think all the treating clusters as cattle, not pets, but there's a related thing that I think really. Everyone should have a development cluster and then a completely separate cluster for staging production. I mean who may be staging a production is one cluster, but they might as well be two really well these new object or if risk is more important than a few bucks with your cloud provider and we don't really have a demo.

E

Yet, how do we make say three clusters one's the development, one for staging ones for production and we completely separate them from each other so that you know staging you can't even get into staging it's locked down and productions, lockdown and developers. Community work in the in the development costs that they can get into pradhan. All that kind of stuff, so I think we kind of need a terraform installer thingy to be able to show here's.

E

What Angus looks like we're doing his excellence like we'd, say three clusters and each cluster could be a completely separate Amazon account or you know, Google account or whatever. They have completely different irons and all that kind of stuff and nobody can cross clusters and we use guitars to believe all the pieces together, which I think would be an amazing demo, because that's kind of what people really probably want to use in the real world.

D

Across different cloud providers will be amazing, yeah.

B

D

E

B

Right so that should be easy, well yeah and Kuban. This uses vows because we're developing apps they're portable across all these clouds, then yeah.

E

B

E

Tricky bit as soon as we start growing multi-cloud, these we're gonna need a Federation to give feedback to developers, but they might not be able to look at inside pod. So look, definitely not like a secrets or conflict Maps, even maybe but developers at least wanna know, like is a pod dying. You know, is a pod up. What's the restart count on pod to be able to kind of get some kind of feedback that.

F

Maybe maybe prometheus is their.

E

Way of doing that- and you just share the Prometheus rather than there yeah Kuban any stuff yeah, but feedback becomes harder when you've got multiple coasters. Maybe kampl can help yeah.

F

Can you hear me just yes, yeah I, think part of the problems with staging production is getting the images around from place to place. So you know, obviously the awsm production is going to have its own docker registry, presumably other than unless we want to go with a shed dock. A registry I've used spinnaker a bit and like Spinnaker's kind of model. This is you installed spinnaker once perhaps sort of in your dev environment, and then that can reach out to different clusters.

F

So it's kind of got like there is our back in there, but it's got an elevated sort of position where it's able to do things to other. You know. So you are the clusters of you, know other environment, so I think whether or not that sort of leads you towards this sort of kind of an asymmetric cluster. Where one cluster has got a kind of elevated. You know the dead. One can do a little bit more walking can tell other clusters to sort of do a little bit more yeah yeah.

E

I I've heard some folks get well soon, as you kind of say it. Dev can push to prod some security people get all kind of freaked out so that you could, you could do push up all right, like I was kind of thinking.

E

We might want to do a pull based mobile so that to avoid dev having right access to the production, docker registry or the staging docker registry, you might want to have a a pipeline in say staging that git has a list of all the approved images, and the first thing it does is: go through all the list of the approved images in some file and gate and girls Oh. Some of these images are missing.

E

Let me call them from the development docker registry and call them into pull and push into the production registry, so that staging is locked down and only jobs inside staging consult the staging and it over production, but you can read from other ones so that you know production could pull from staging the polka various registry. So we could have these. You know one job for the environment which is allowed to move things.

E

Maybe you want to move them out of balance, possibly I'm, not sure, but I figure, something simple like that, might be a nice way of at least meeting each environment like staging production can be completely locked down and nothing can write into it unless it's apart from itself and then it pulls everything from there on then you've got a nice simple security policy, but there's that we could do either we can push and pull.

E

The worry is a you know a pipeline pushes by accident before it's been approved, so you can I kind of like there defusing get-ups. It's the only way of getting images into production. Is it's a pull request and you can have courage of you and every pull request to go I, don't think that image has been approved. Yet you know that image hasn't been properly tagged and it hasn't been through security checks.

E

You could give the CI job in production could fail on the public request if the angkor security scanning hasn't validated the image and all those kind of things, it's got three plus ones from the production, ops, folks or whatever. So I do like the pull model. It's easy to do guitar Swift, but we could do either though I mean people can choose different way. This is just a pipeline right, so people can they were like yeah, good question.

B

um A demo of something in this people have more questions. You can just shut them out. Yeah. Let's do a demo who's gonna go first, you wanna go first or me, which is better. Yours is probably getting better. Now that you've been hacking. Last people I.

F

B

I've got icons, so maybe you should go close. Yeah I got link.

B

Note to self next time tell James about the after the demo.

B

Okay, all right! Let me share my screen so.

A

While you do hello everyone, my name is Liam Newman I'm, just running the host on hosting this and on the hangouts we're about halfway through here there's, uh obviously, if you're you're on the participant link here, you're already, you already know about this, but on the slack Channel for Jenkins X, there's a link to actually participate.

A

If you want to be able to talk back, you can also ask questions in there and were will be taking notes in pulling links out of the chat window and stuff like this, so that it'll be in a Google Doc when we're done, which I'll also add to the youtube description. When, when, when we're done with this chat,.

B

Wolfson thanks the.

A

Matrix: okay, okay,.

B

Alright, um so the new cool thing, so let me give a bit of background myself, James and Webber over Q Khan two weeks ago. I think it was now in Copenhagen. Excuse me and we got on the Friday. We got to catch up with some of the folks in Microsoft and they were showing us some of this stuff they're doing around vs code extensions, and they got some really really interesting things. My team just mentioned in chat around also integration. We say the cuban sea service brokers, so they can.

B

You can actually on from the IDE from the vs code. You can actually look and find discover services from internally your cluster or from the service broker and it'll generates the actual service linking and the environment variables you can use to copy into your code and she works best.

A

We're good I'm gonna hop in here real quick, sorry, yeah, take your whatever you think your your font size is and pop it up until you until the point at which you say: oh geez, that's too big! That's how big your font needs to be. Okay,.

A

Maybe one step down from that.

B

Okay, so um so kind of got us thinking, we've Jenkins actually have the CLI, which is, but he also now be pretty happy with and we're using it day to day jobs as well on Jenkins X project bill, Jenkins X, one of the things we were looking to do is actually have some kind of experience within the IDE.

B

So we've created a new repo vs code, JX tools and there are pipelines all set up, so you can actually bend test fill pipeline before vs code extensions, and you can actually fork this and contribute back to this. Any contributions would be amazing and we'll also bring things get merged. You also have pipelines and then actually publish that out to the the marketplace, so they can appear in vs code. So it's pretty basic at the moment.

B

What we have at the moment- but this is the- let me just give you an example of what this is very quickly. This is the project. The vs code tools, Jack's tools when you open this up in your vs code IDE. You probably can't see this, but on the left hand, side there's a debug button. You can start this debug and what this does is actually creates you a brand new vs code window.

B

Now it's trying this bit bigger. What we're gonna do now is James is going to talk you through this bit on here. This is a Jenkins X window down here, but we're going to add. In a start, a watch JX watch activities. Well, hopefully, what we'll do this out of the box, but what we ultimately want to do is start being notified of when activities of a pipeline happen for our applique project goal and q1. So we just this is going to be a pretty rule demo. So what I'm going to do is.

B

Start the pipeline for my goal: ur by go project, that's going to trigger a job in Jenkins. You can see if anybody is familiar with with the JX activity get activity. This is actually watching. The activities uses custom resource definitions within kubernetes to the CLI. There's a CLI command, J X get activity, F go along, you one watch, so that's going to watch using CLI. We see, we've got some activity now we have everything here we can see now at the bottom.

B

What bottom right-hand corner we've got some notifications now that there's probably a few going to be coming through here at the moment, maybe we'll turn these down a little bit and but what we've got now is we check out the source code it'll build? We actually get a notification when this application version is deployed into a staging environment, actually has a link as well then to open the application from the staging environment. So you can open it up from your browser, so you're just staying within the IDE.

B

It's just something we're experimenting with at the moment, trying to give a richer experience of you know just not necessarily having to leave your IDE when you're actually doing development, and it probably takes a couple of minutes. You might not see this now, it's probably probably worse and back to James, but this is just wanted to show some of the things we're kind of exploring with vs code extensions and then syncing three with the abilities with with Jake Jenkins X. Well, we've got a build release and eventually, we'll have a link that comes through.

B

That was really it. I just wanted to show there James you wanted to do a little bit more on that.

E

As well yeah, that was also I'll.

B

Step sharing, let me see yeah you see, focus engineer. I was just hoping to check it see if that last one pops up not know enough time.

B

E

You see my screen, I need to click on. You I think maybe turn off my video turn up my video there. We go I'll telephone video just in case my Internet's, pretty dodgy right I've got a little there. We go there's my command line. That's watching an OGS thing is that the build here, so this is via scurred I'm running this in debug, because I'm not quite checked in all this code. I've just been working on this little folder thingy, so you might have seen these folders in this code.

E

So there's a kubernetes one that lets you browse cube annuities, but this is my current namespace and that when we made this bigger, so I can look at all the current workloads and pods in my namespace and those kind of things. So the cube innately stuff is pretty cool. So we've now got a Jenkins x1, which is basically browsing a real-time view of all of the repos and projects. That's inside check is X right now, so here we've got a bunch of different builds and, if I trigger a build, let me do that.

E

You know it's. It's see that little icon, we have a little icon because build six has just started and you can see both started there so we're watching in real time all the builds that happening in jiggies X. So we get a nice little animated icon, but shows the builds running. If you click on it, the tooltip tells you what's happening.

E

It's been six that won't succeed, though the status the real estate isn't terribly useful right now, I'm hoping to put a bit more information there and we can't yet expanding slide the build to look at all the different, all different stages, but we can at least go do things like open the pipeline log. This opens a browser window.

E

We can open the use of your pasta tree nice I'm, not working. What's the browser with the gun? Okay, that normally works. That was working like 10 minutes ago, probably view the pipeline. That's not working either. What's happened, I don't know. What's happened to my browser that was totally working.

E

Oh, let's try this one. Let's try this one. It.

A

Might be opening.

E

Off so that so I can look at it's just. He doesn't have that URL. Yet the pipeline look so there's the pipeline log, so I can look at the log of this old pipeline or I can go to the Jenkins page to look at the pipeline details. So it's just a simple little way to start with a browsing what's happening in your cluster, so we're hoping to tie these two things together so that you can either browse everything.

E

That's happening in detail using this simple UI to see what's happening and we can probably do watch the browser log in a terminal to avoid opening a browser window which might be nice, and we can do more more actions than this like start, build, stop build and various other things like that and I think tying it into this stuff. James just showed where, if you're working on one of these projects, you kind of want to get notified when something's happening on your project, plus we've got this Explorer to look at all of the other projects.

E

So here's my environment, for example, we've got little girls. There's a few failures, don't know yeah. So that's it pretty simple stuff. So far, yeah.

B

Actually, if you just switch back I'll, just I'll show you my other thing just worked. Actually, if you don't mind yeah, and then we need to figure out where's the stop right at the bottom. I think it's oh I.

A

Can actually run that a bit you can go ahead and take it there, you go okay, thank you and.

B

You can see.

A

B

Down on the bottom here, we've got the extra notification, so my app has been floated to stage and I guess the application here. So you think that, and there we go our amazingly colorful demo app. What we should be able to do is add in you can actually add actions on to those notifications as well. So perhaps we want to look at.

B

Perhaps we do want to add some kind of promotion, maybe on that as well. One thing, though we can actually, this is all in the in the marketplace at the moment. So you can actually go to the moment is beat is definite preview at the moment, so don't expect too much and maybe just nothing more than what we're showing at the moment, but you can go and give that a go and test out itself and maybe try hacking on it as well. I think it might be pretty cool, developer experience and yeah.

B

Yes, I think that was just the things I heard anything more from you James or was only comments or anything any questions or any ideas from anybody. I.

E

Wasn't going to mention how Microsoft's been doing some really good, then I was like well. What am I excited about? He is killed at the moment which everything eyes he starts off crazy, quick. So it's really good for micro services and just opening project and hacking on it.

E

But the thing that God is quite excited was the this stuff Microsoft's do the s-code is looking really awesome. So they've got the bugging working in most programming languages already in pods in Cuban eighties there's already, some Cuban eight is tooling, which is pretty reasonable, so you can browse resources and stuff, but there's also increased tooling around things like helm, charts and using the service catalog with ham shirts, which is looking really exciting.

E

So, if you're working on an app that uses of database- and you want to use a service catalog to find your database Keys find your database tokens and whatnot to access the database they've got some pretty slick, looking UI for that, so I think fairly soon, I think editing, debugging and browsing, see ICD pipelines and working with helm and Service. Catalog should hopefully be mostly an IDE automation things. So people won't have to understand all of this magic, yeah mom and go templating and all the kind of the magical recovers.

E

I mean it's not quite there yet and they need a bit more polish, but I'm, hoping in a few months we'll have a really slick, give up experience with BS code. Where you can work on traditional, vanilla cuban it is you can work on how or you can work on jenkins x or all three of them. We, though, without the service, catalog and I, think.

E

Yes, so I think the S code is going to be pretty awesome. Developing microservices and covenants.

E

Yeah blue ocean, those graphs and visualizations of whatnot, so we're hoping not all of the links right now in the BAS code plug-in talk straight to the ocean I'm hoping to fix. That's all the nice blue ocean screens appear, but we could look at doing some BS code, specific visualizations in vs code, so we don't always have to open separate web browsers. We could do some simple visualizations find the s code as well get time.

B

We were gonna fill it fill the hours 15 minutes, it's not too bad. Oh there's no question. Think spinnaker has some visualizations pipeline progress right, okay,.

E

Yeah and and blue ocean has got some pretty nice visualization of pipelines themselves. I. Think one of the the challenges there is that the visualization Jake is tends to be quite generic.

E

What we kind of want to do with Jenkins X is a bit more show like in many ways a build is just a build and we don't really care much about the detail too much and honestly fails, but we mobile do berries, what are the environments you're promoting to and what are the links of environments and where's the pull request for the promotion, where's, the pipeline, the environment, promotion and so on. So we kind of want that kind of a bit like the command line shows when you do JX.

E

Let me show my screen again when you do JX, which I feed you off. Basically, when you do JX get activity and you get this now. This is true and it's looking at wrapping on them. This is almost showing you of the edited highlights of what we want you to know about, like really. We want to know what versions being released. What's the links of the version, so I can look at it in say, github or get here? Oh good luck! Where are the pull requests?

E

And then, where are the actual applications running so I can click on the app. So if those link between either the pull request, the version, the CI job in the promotion and the actual app. So it's not just having a nice box diagram, it's kind of one nice useful link so that you can drill down into that detail, maybe the best way of doing that is through the tree which widget. Maybe we just expand that in here, so that you could expand a build into pipelines.

E

Here we could look at the promotion and whatnot right here. Possibly we might want some kind of UI to visualize a pipeline a bit like the ocean, but maybe slightly more developer. Centric. Yes,.

E

Let me see if, if you figure to stop cheering, oh there is I felt good phew.

B

Cool and so how was that being fit for people? There is this being useful, I mean people stuck around so I'm, hoping that's it's being a. We have a rough idea of what we're going to talk about, and sometimes it's just good to. Let it flow. But people asking questions is doing a feedback so far.

G

It looks awesome, I'm really excited to to see how quickly everything's moving.

B

Awesome, yes, we like to move fast, but then there's also a lots of opportunity for people to get involved as well. Well, there was I was talking to somebody over. There was a blog post or something we were chat. We were having some day and it was just amazing to see the community that's building around Jenkins, X I think he's just going exponentially and I think that's truly amazing.

B

James you put outs and we've got these metrics that come out once a month or something different contributors of people getting involved in the project, whether that's from issues or fixing pull requests. Even just sharing I mean the slack channel itself. Is people are helping each other and I think that's brilliant. This is just such a short amount of time and we're really hoping that this is kind of a kind of start. It starts the bacon, more people get involved and it kind of evolves organically, like this. So I think this is the start.

B

This is the first attempt of this live hangout, but there's gonna be lots more ways to be involved in the community. So if people are actually viewing the recording afterwards, I just encourage you to jump on slack as a ninja. There's a start and it's a really welcoming place: Jenkins X, dev and James X users channels and have a look at some issues or even if you don't have that, don't know, go long or don't know typescript for BS code. Extensions, even just sharing ideas, I think is a great way to start off.

B

E

We could do with more ideas on the v8 code. Looking particularly like we've only just started really or sto awaits is another thing: we're desperate before yeah. We want to treat STI user groups thanks Jason. We really want to do environment promotions, but not right. Now we use name spaces or, and all clusters we'd like to be able to do multiple promotions across environments in one cluster so and one namespace.

E

So, for example, beta testers on do a canary first then do beta testers then do employees, then, do you know 10% of the rest of the users and do running upgrades across users. Cuban areas already has running upgrades, which is good. It doesn't yet to automatically do you use a base rollouts, which is where each tier comes in and things like, ambassador and stuff, so I I'm, hoping we can get that in fairly soon, because I think that would be really really useful. Yep there's lots to do, which is good. That's a good thing.

E

It would be really boring if everything was finished.

E

It's awesome love to do.

B

Yeah cool all right then well, if, if that, if there's any last chance for any more questions or any more comments or anything.

B

Okay, well maybe we wrap this up in there, we'll we'll have them another one in in two weeks time um see you all online thanks, so much for have been involved. Okay, thank you.