Red Hat OpenShift Developer Experience Office Hours, 25 Aug 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Developer Experience Office Hours: CodeReady Workspaces

Description

This week @natalevinto and @briantannous will be joining @ChrisShort and @ryanj to learn about using 'devfiles' to configure Eclipse Che / CRW

A

B

So you don't have.

A

To write it from scratch, uh which is good right, uh we don't have the uh we don't have to do this, so we can just look and how it's generated.

A

uh If you remember there was that if you can see the project section so the name of the project, then, if you look at this, it doesn't start from git uh starts from a zip file contain it inside the openshift that file registry. um We have a list of components for, for instance, the plugin, the visual studio called plugin for typescript, then the node.js, the back, uh and we have our docker image the container image that is actually running our uh commands.

A

uh If you remember, we have two containers, so we have this container that will have some environment variable here and the run configuration. So we can also code this run configuration if you want to give to our team. You know uh the run configuration command to just try start push deploy. We can just act this one and create uh the that file for for for this one, um so as an exercise. Now what we can do is uh you know, adding some plugin to the to the to the workspace.

A

For instance, if we want to uh also from this view, the open shift connector plugin- and this will contain auto, for instance, uh open shift cli. So all the plugin that we we need, we can just click apply here. What happens? The the workspace is going to be stopped and restart, because you need to mount the new plugin in the pod. It's like a you know. A docker volume in kubernetes is a persistent volume.

A

uh You need to to mount this and uh it will uh give us the uh openshift connector plugin visibility uh when it started.

A

So um this is nice because uh from the ui we can do everything uh as we want, but if we want to let our teammates to don't do this manually all the time right and maybe start from some git repository and not starting, uh for instance, from a zip file, what we can do is just copy this dev file in our kit, repo and then starting, creating the new workspaces from the uh quadratic species button that I can show you uh in a few moments.

A

So their idea here is just to create a code for our developer. Let's say we want to put in this organization here we want just to put uh this mongodb example, and this mongodb example itself is part of a a list of example in uh eclipse check project. So you can look uh at all of them and if you go here in samples very easy on gizzard who.

C

Has the dog, I'm sorry, is that unitary, okay, no worries, I'm just curious.

C

A

C

A

Very big dog of my uncle that is usually start singing at five six, for I don't know which reason maybe my voice, so I'm sorry about it.

D

Oh no worries I'm. I was just curious yeah as long as that's the person speaking then uh yeah. Nothing can be done about it really.

C

D

Than take the dogs out, no, no, I don't bother with.

C

That yeah, like I'm, used to having dogs around right, like my dog, is right there. So yeah right! Oh.

D

That my dog, my dog, goes off every day. Five or six o'clock starts barking and I I gotta go, take him on a take him on a jog or something um he's predictable, at least but uh yeah no big deal um hey what, since we paused, we had a question in chat about setting memory limits on uh on some of these environments. I think you.

B

D

You had a two gigabyte limit on uh the development environment. I believe is that.

B

Correct so you can set the limits uh even.

A

You can set the limits uh in the dev file itself, so, for instance, in this container here for node.js there's a limit of 512 megabytes. Those are relevant to the workspace to the container right. That's.

C

For the workload right yeah, that's the workload! What about the.

A

At the upper level and openshift um you can you can set uh quotas uh for the specific uh openshift project, so this can could be done either from the custom uh from the custom resource uh when you style coded workspaces. You have those custom resources here, I'm not administrator. Let me log in so I can show you the.

E

Yaml that you showed for code ready workspaces earlier, so we're trying to yeah.

A

E

We were saying yeah yeah so that you can set limits on the amount of like workspaces that each individual user could have open at one time so that you know you're not consuming too many resources for each individual user. uh You could do you know other things like memory limits for that environment yeah. Usually this type.

D

Of question, I would want to ask the person like okay. Well, why are you trying to limit.

E

D

And who are you trying to limit it for? Are you.

B

On the operations side- and you.

D

Want to restrict too much usage by developers or give developers a sensible uh ceiling that they shouldn't go beyond so they don't have memory leaks or you know, or maybe you're the developer. Who is trying to prevent those memory leaks? I think they both need to have uh entry fields uh for folks. um The ones that are more uh limiting per project are probably uh what I might use for administrative on the administrative side.

D

But a lot of this, I think, is slightly a lot of what we're showing today is related to the setup of code, ready workspaces and so a lot of kind of what we're showing is a little bit more on the administrative side. I think I think a lot of this could be reduced down if you were a average front-end developer, who wasn't really focused on containers or kubernetes. um So.

C

D

Boiled down quite a bit more.

C

Jp dade is like right in my heart right now, because he just said yes developers trusted but verify I'm the ops guy. Those jokers will take it all. If.

D

C

That's the motivation.

D

C

Right so now we know right, like there needs to be some limit on the amount of capacity that the regular you know, user can uh consume as they're trying to break things right like oh dbas, is a good example right, as they load in terabytes of database into memory. Potentially right, like that's fun, to deal with as an ops person. I'm sure right like there's. Is there a way to limit that.

A

um Yes, you, you can limit, uh as we were talking about uh from a operator perspective or you.

B

A

Just give quotas to the project right.

B

A

For instance, in our cluster here there are uh some some quota, uh not in this case, because I think uh I removed it, but you can also set up a trigger or uh on our web book admission control and a web booking open, shifting kubernetes uh that set up limits range uh or the resource quota in the new namespace automatically.

A

So this can be a very helpful if you have a very large number of developer team and you want to just assign quota or limit uh the the amount of resources that they can use because they could hack the daf dev file, but they can't really act their limit range or their support. Yeah yeah.

C

That's that's gonna, be your answer jp date. For reality, all those problems right like.

A

C

You're gonna need that on every project, right like that you're handing off to somebody or every name, space that you're handing over to somebody right like just like you should have resource limits in every container. You should probably have resource limits on things that you're handing to people as well.

A

Exactly exactly um the code ready workspace class customers is very granular in.

B

Terms that you can.

A

Say: okay, you can the dog started again.

C

I'm totally fine with that. I'm ignoring it at this point so.

A

C

A

Okay, you can use just uh two works: two persistent volume uh per uh two per system- volume per user and the sides can be maximum two gigabyte per persistent volume. So those are settings you can put in the yaml file here and you can drive uh the the tuning of the convertible spaces instance on the storage side, for instance, um but this is.

B

A

Administrator part.

B

A

If we come back to the developer part so as developer, if we come back to developer parts, what we want to do is I want to give to my teammates. If you go to these check samples, you can pick one of this. For instance, the one we selected is this node.js mongodb sample. I want to give the capability for my teammate to create this automatically, maybe with the openshift plugin connector, without doing it manually, so how to do it? You can fork it in in the in your organization, so I'm gonna form this repository here.

A

uh It's gonna be forked in our redwine software organization, and what I will do is create is adding a dev file here so code, ready workspaces with a mechanism which name is factory, uh will create the workspace automatically just clicking on a on a button, and uh I can show you that it's really nice, it's really easy. So if you remember, we have just to go back to our dev file.

A

Here uh we have just to change uh the the source to uh type to git and not using uh you know the zip file, but everything then uh would be would be. The same will be good, so working in our fork. uh We just add a new file from the by itself that file yammo. So it's a yaml file. We can copy the yaml file here and we can start modifying it.

A

So we say that this is git and we want to start, for instance, from this repository itself, so we can give that url here of this one. So let me let me open this and we can just copy this one.

A

So location is going to be this one. So we start with the project. Name is going to be uh project something, so the name would be proud: node.js, for instance, um office hours, for instance, we can just.

E

So if somebody is doing this yourself, I mean obviously make sure initially, whenever you load code, ready workspaces log in and get the sample, um you created a node.js sample, so create one. For you know whichever language there was all those choices right so start off with that and then you'll be able to easily switch out. You know this over here right.

A

Exactly exactly so, once you have the death valley here, uh the nice things is to just add in the redmi. I just click button that will start this automatically for you. So in order to do this, uh we can just copy. um We can just copy another that file sample uh and we can- and I can show you, for instance, the one I was doing uh for the same example last time. So, let's go here. um Let's go to my wrapper. I already did it: uh let's go to the the same example here.

A

So what I did is just adding this button here when you click here. This will create the workspace for you using the dev file and using the factory uh the factory set. So I can just copy this one. No uh coming going back to my example, here put it somewhere- or we can just put here, for instance, um and what you have to change is uh you have to upload this factory svg, which is the image or we in for this example. Here we can just give that the url doesn't matter.

A

We can just copy the url address here from the other repo doesn't matter, and but what you have to do is give the code ready workspaces endpoint, so we have to copy the code ready workspaces endpoint in our case, is uh this one, so I'm gonna copy here and you have to give also the you have to point to this uh rest: api endpoint, which is factory accepting the url of the repo containing a dev file.

A

In our case, we just copy our repository here, so we have just to go here and um copying the repository. Let's go back and put the correct url here: https, okay, so this is our code ready, workspaces.

C

A

File, uh we want just to click here and start a new workspace in our user uh um f4. For so, if you imagine one of this teammate chris orion or brian, can just click here and when you click here uh this gonna, if everything is fine with the syntax of the dev file, everything is fine. With the uh configuration of the related workspaces, this will load the factory endpoint, uh so the factory endpoint will show you that we look at the dev file, uh initially the workspace and start the workspace.

A

So here I'm not allowed to start more than one workspace, because the administrator didn't allow to do it for me. So I can do two things. I can relax the settings from the custom resource or as a restarting the the application, or I can just stop this one, because I don't need it anymore uh and I can let the other one start for me. uh So we can just go back here and retry.

A

We are logged in here as a normal user. So we can see the administrator developer view yeah. Okay, that's fine! So we can start again. So we we stop the the other one. So we can restart again.

A

uh If there is any other settings, we can just relax this on the custom pressers we can put that we can start more than one workspace uh and what happens under the hood again is that you create a pulp in your server when this starts and uh when the the spot start, you can have the quadrat the workspace, starting in your uh in your environment for your user. So this is going to create again a new persistent volume.

A

uh This is gonna create a new persistent wall for us when the persistent volume is okay, uh it's gonna create uh the the container for the workspace. So if we come back to the code ready here now, it's it's going. So um this will load our dev file with the openshift connector plugin, with the same mongodb and node.js call so the new my teammate can start working uh using also the openshift integration with code redicus faces.

E

So if we yeah, you could see, you probably can't see natalie, but everybody else could see. uh So I went to that repo and clicked the button and mine's also coming up now under my account, you clicked the button. I clicked the button. I clicked that that button so that I could create this workspace wow.

A

Yeah yeah and if now we have a very uh limiting settings um for the workspace, so we can relax these settings. I did it for, for instance, for a demo with a mini cube and eclipse chair, which is the the same thing, but using mini cube, so you can relax these settings in the custom resource and you can say, uh custom chair uh properties, limit user workspace run count.

A

If we do this in our workspace and if anyone is not using that uh in the while, because the the chess server will be restarted, then we can start multiple workspaces per project. So if you see we can drive as administrator or the limit count of the workspaces and give more freedom to the developer team, if they need to open more workspaces- and usually they do no, they want to test, they want to try. So now we started the workspace. If you remember this is the our quadratic workspace.

A

The code is the same, but this time the code comes actually from the git repository. So we have a git view here. If we do any change, we will have the possibility to review all the change we made with a you know: live div everything that uh normal modern id is doing uh very cool, that we have the openshift plugin integration and the kubernetes one. So we have a open shift, plugin integration.

A

uh This means that here we can create the application in openshift we have the kubernetes one, so we can navigate uh navigate all the kinds and work on that, uh so we can really deploy this application and make it live uh that linked before was just uh living inside the workspace.

A

If we want to push inside open shift, we can just doing from here it's going to create the node.js application in our project in openshift and and then we can uh start uh deploying the thing so each the nice thing that each developer me you ryan, chris uh brian, can uh work in its own project. So here I can work in my project. Ryan works in his project, brian chris working in their project. When everything is fine, they test in their own project.

A

Maybe they can use the detect on plugin for quadratic workspaces, so we can have a tecton plugin here, a text on pipeline pushing the application to the to the platform. um So this is gonna, be the the flow. So now, if you want to, we can deploy the application inside open shift or we can go through the uh talk about audio.

A

I think the nice follow-up here is that uh when you use quadratic workspaces uh plugins for uh openshift connector here, uh so you have the connector. The connector give you also a terminal inside your uh code, ready workspace. So here I have a terminal which is different by my node.js mongodb terminal, and I have the audio so auto, which is the command line for developers pushing for pushing application inside open shift.

A

I have the oc command line, um so I have all the uh settings uh that I need I need I I have also took ctl. I have all the settings that I need. Only thing uh I want to do maybe is login to openshift classes, to the right with the right context, and I can verify that here, I'm the right user uh and I can maybe create a new project uh where to put my application.

A

So this is uh uh very simple. You can also mix the strategy for, for instance, this repo has already a db deployment file that you can deploy. So you can use this strategy to deploy an openshift or you can just deploy from much much better and easier. You can deploy your mongodb instance from the developer. Catalog, it's straightforward! You just go here. You pick the database in this case. It's going to be.

A

You select, also persistent database, and here you have to put all the settings needed by the application.

A

For this application itself, if we want to see how it's made, we can just inspect the code, we can go inside the code and we can parse. um For instance, we can see the uri here, it's user password one would be very simple no, but it's just for uh for for an example.

A

So if we want to create this, we can I we can create uh the database like it is expected there, uh and when you do this, you can have your uh mongodb database ready to host the apple, uh how to serve the application that you're gonna actually push via open shift. uh Plugin connector.

A

This is a nice follow-up, because now we're creating through our auto wrapper ui, which is the openshift connector our application. uh So we can just create the application, we call it the guestbook.

A

We can let this application start from the workspace directly. So we can start from this code ready workspaces workspace, so we can tell okay grab the code from, for instance, from the mongojs, the node.js mongodb example.

A

uh So this means that audio will go there and try to deploy the application, but we have to instruct extracting and give the name uh of the component and and the programming language. So we just tell to the openshift connector. Okay use a this is a node.js application.

A

Please push it to open shift when you push the application to open shift. This means that here you're going to create your application and that this will connect to the application uh you can. You can do some grouping here for the application itself. You can play with the arrows and connect things when, when the node application is ready, we can also access it from this plugin here nice that you can also create an openshift route. So you can access the application if we call it in this case, just book route.

A

Remember that you have always to push your change and oh because auto will push your change. This is a wrapper ui wrapper for auto. If everything is fine on the application itself- and we can review this by logs if he could connect to with the right settings, uh for instance, if the connected to yes. So if everything is fine, this is going to be our application, which is available not anymore from my workspace, but this is going to be available from my team in the cluster.

A

So this is our application that I can work in my openshift environment, so I tested locally to the government workspaces. Then, when everything is fine, it's pushed automatically with the openshift connector to to to open shift and thanks to that file. If you remember thanks to that file, I could automate this and let the openshift connector be the default choice. So when chris brian ryan, anything in this red wine organization want to start coding, they start from my lightest change on the code and from my dev file using the openshift connector.

C

So did anybody put anything in the gas book yet.

A

Let me check if anyone put something in the guestbook, I can put something. uh I can't click here. Let me okay, let me say: okay, hello from openshift, and here we go now great message from chris, uh so yeah our inner loop, no work it out very good uh because we tested locally in terms of uh local workspace.

A

Some good ops mate gave to give to us the possibility to create workspaces. uh Persistent volume have some storage and have the code ready, server the multi-tenancy, and so we implemented this developer loop here, and this was for that file version one. We know there is the file version two and I'm gonna pass the ball now to ryan. uh That can talk to us about the version two that can be introduced by the new version of audio you've.

C

Been 11 minutes and 30 seconds ago.

D

11 minutes, oh man, you know.

C

D

We have a couple questions in chat as well um that I wanted to get to and uh we might have to save some of the dev file 2.0 for a later uh first follow-up meeting. um I can definitely give you the uh a recap of what I tried uh last night uh regarding odo 2.0 and dev file 2.. um But let's.

B

D

In on uh on chat, real quick first, um so we don't lose these uh questions.

C

Yeah so there's a question around uh persistent volume groups or persistent volume, and you know the containers for pods ncrw uh how many are needed? What's that look like.

D

Natalie, do you know very much about how many pvs are used uh per crw session? Is it generally one pv per session or is a lot of that backed up in the dev file registry? Oh so.

B

A

Couple of pv use that and we can, we can show that. Actually, uh if I can start sharing my screen again so jp.

D

Dade was mostly interested in figuring out budgeting per user. I think he has a hundred persistent volumes available. I don't know how many developers he plans on onboarding, but if.

B

D

Uses two or three pvs just on its own.

B

D

Then you know your your database might also be using a pv. This mongodb example we have um just the workloads themselves, may need pvs as well so um yeah. I know I was successfully able to go into your github example: click on the link to spawn my own uh hosted workspace, it booted up, and I ended up at first time. It failed because I hadn't accepted the um invite to the red wine software github org. um So I.

C

Had to go back.

D

That was on me. I had to go back in and look at. My um group invites approve that. Yes, indeed I'm a member of this github org and then I was uh allowed in to spawn my session so security's working, which is good, um and I was able to just directly one click boot into a hosted ide that also had a terminal session in there, as well, with git and odo and um most of my tooling. So um thank you natalie for successfully onboarding me as a as a junior dev that that all worked.

D

B

D

Yeah, well, you had to do uh a decent amount of setup live on the stream. Today we skipped over a little bit and that's partially one of the other things that jp dade in chat was asking about um what else.

D

What other work is involved in successfully rolling out this experience to a team um because they're, uh I'm not sure if our current documentation goes as far as connecting all the dots, um quite as well as uh what you've shown today um to the point where all they have to do is just click on the github readme and they're in business. Right.

C

I think that, like setup show is a great show right, like that. Whole process could be a show all by itself, and then you could incorporate dev files v2, and we could do that next week. If you want right.

D

C

D

You're going to.

C

Talk about it right now,.

D

We will yeah, we need to check in with serena. I know serena is I'm not sure if she's around in chat she had. I think, uh first rights to naming the topic for next week. Absolutely.

C

Excuse me she's a future. She.

D

Knows best she.

C

Says I am in next week.

D

C

I don't, I don't want to stomp.

D

On if she already had a topic in mind, I don't want to overrule whatever she had in mind. We've.

C

Got I mean there's like two or three brewing right now, so yeah.

D

B

C

Serena's been watching and we can.

B

D

For sure, but we're not stepping on it.

C

Serena's down for the cause right all.

D

Right well, yeah! Well, I don't know what, uh if she has things in the road map that it's like hey, this is about to roll out, and I need to talk about it at the start of september.

D

You know she's.

C

From the future, and now I know that.

D

I don't know I.

C

Know, but serena also knows that she could get like future times for me if she wants to.

D

C

D

Right well uh definitely stay tuned for follow-up episodes. um One thing.

B

D

I can potentially demo- or I would challenge you to try out on your own um if you're up for it, if you're interested in learning more about dev files, I posted a link to a github issue at the very start of this, and I'm going to.

B

Repost it in chat.

D

um That has a kind of the dev file, 2.0 um roadmap, uh the url, it's github.com, dev file, api and then milestone 1 was the rest of the url there, and so that has kind of. What's going into this dev file, 2.0 release, I know odo 2.0 has uh been released. I think this or there's like a. Let me double check on that. uh It may be in an alpha state. I think the latest one I grabbed was odo 2.0 alpha.

D

Yeah, it looks like we're still on alpha releases for 2.0, which would make sense, because the the dev files are still kind of in progress for 2.0, but um odo 2.0 is targeting these uh dev file, 2.0 format, formatted dev files, um so that's kind of uh what we're working on out into the future. Some of that ui wise uh won't be available until I think open shift. 4.7 is where it's currently slated uh feature-wise.

D

um So um if you want.

C

To play your projections or small yeah, all future yeah all subject to change, but if you want to get.

D

A small taste of the future- this is what I did last night I went into um catacota. I picked the top right um right. My video flip, my top right option was developing on openshift and then the second after that you get into a sub menu you can pick developing with odo is your second top right selection in there.

D

That will automatically drop you into a catacota scenario that helps you get started with using odo for the command line, um it's different than using code ready workspaces. um But what I did was I went to the releases page, followed the instructions for the odo release copied in the latest odo binary.

D

So I had odo v2 and then went ahead and followed along with the rest of the um openshift catacota learning scenario, and it all worked uh perfectly. There was one extra command line flag. I needed to add in I needed to add in like a dash dash s2i on the java build that was a new flag. That's been added, but otherwise odo 2.0 the alpha seemed to work fully backwards, compatible with odo 1.0.

D

A new feature I saw was that you can configure odo 2.0. You can run, I think, odo registry ad and then you can add a dev file registry address and then odo will try to work against that dev file registry to load, store, save and interact with those dev files.

D

I was almost expecting one to get output to disk or something I didn't see that, but it almost kind of makes sense architecturally to keep the ide configuration somewhat outside of maybe the source code, especially if I'm dealing with microservices. I may want to decouple my ide set up, especially if a lot of people tend to configure that per user.

D

My ide setup might not be your ideal ide setup, so I don't know if it really belongs in a in a github repo, but it's definitely nice having a link right in the readme that I can click on in order to load uh an ide and then kind of hopefully merge in some of my preferences as well.

D

So thanks again, natalie great demo today and we'll be back every week um same time same place uh thanks for joining us this week. uh Thank you uh to the folks in chat who had questions um and we'll be back. Next time, as well, anything else to close us off with chris.

C

Nope stay tuned for openshift comments, starting in 25 seconds. All right. Thank.

E

You all so much. Thank you all right. Thanks.

C

C