GitLab ⛅🌱 Cloud Seed, 16 Dec 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 11. #everyonecancontribute cafe: 5 min prod app with Vue & CI template insights

Description

- Repository: https://gitlab.com/gitlab-de/5-min-prod-app-vuejs
- Blog post: https://about.gitlab.com/blog/2020/12/15/first-code-to-ci-cd-deployments-in-5-minutes/
- Deploy template development: https://gitlab.com/gitlab-org/5-minute-production-app/deploy-template
- Live from AWS: https://twitter.com/dnsmichi/status/1339201532416372737

CI Insights

- Includes: https://docs.gitlab.com/ee/ci/yaml/includes.html
- Built-in templates: https://gitlab.com/gitlab-org/gitlab/-/tree/master/lib/gitlab/ci/templates
- 3 YAML tips for better pipelines: https://about.gitlab.com/blog/2020/10/01/three-yaml-tips-better-pipelines/
- Security Webcast: https://gitlab.com/dnsmichi/ci-security-webcast-2020/-/blob/master/.gitlab-ci.yml
- Tests with the same job name: https://gitlab.com/dnsmichi/ci-cd-playground/-/merge_requests/6/diffs
- Anchors und extends example: https://git.netways.de/packaging/pipelines/-/blob/master/rpm-base.yml

Learn X in Y minutes: https://learnxinyminutes.com/

A

And now looking into the screen preparing live stream, and I think we should be live, yeah we're live now and we can start talking or I can share what I've been doing all day, long about the five minute production app and things around ci cd. I hope that this is okay. We can also try out like matlab or ci rules or ci things which we learned like yesterday in the the kid lab teams.

A

Kit lab ci team asked me anything session. Yeah, I'm open to everything. Basically um it's it was just um one story for today I wanted to share was um just like trying things out and see how things work um and just for context. um This is basically a longer blog post around the five-minute production. App um the too long didn't read. The storyline is um when you want to deploy something to production oftentimes, you might not have a kubernetes cluster.

A

You have might not want to use auto defaults or some like opinionated ci templates, um making the entire uh deployment very complex oftentimes. You just have a website or like something a web application with a database backend- and you don't want to think about. I need a virtual machine for the web host. I need the host for the database. um Maybe I want.

A

I don't want to take care about schema migrations and things like and everything around this and since um we've been working on making it easier, not only with the other devops templates, but also with autodeploy. For for aws.

A

There was the idea to like say: hey give us your repository, give us a docker image, a docker file, and we do the rest for you. You just include a ci template file, which should be automated with a single button in in the future, provide us with credentials for aws.

A

So this is basically three ci variables we need to define in the future. This should work automatically with aws or oauth and providing these these details, and once everything is done, the magic starts, and I will show you the magic uh rather soon in in a live example.

A

um The main idea is that you built you're building a docker container you're running a terraform apply, starts up the resources in aws later on. We might use different cloud providers for that, but terraform runs the plan and then applies the plan.

A

When you have a database host, you automatically get the. I think it's called the t2 micro instance, which is in the free tier, so you can try it out and it doesn't cost you anything you can even let it running like.

A

I do because I forget about things um and then uh basically, when the when the resources are being provisioned, um the docker file or the not the dockerfile, the docker image is pulled from the gitlab branch history, which is already there and it's it gets started and one cool thing, and this is something which is not part of the blog post.

A

Yet because this was published yesterday, um there is uh ssl ortilla support with let's encrypt inside, so you just need to provide a an email address for this for let's encrypt and it automatically gets a tls or ssl certificate. um The other part is the the nip.io domain provider. If you don't have a domain right now, just for dns entry and not using a domain, and there is more more like having a review in my environment.

A

Similar to the review apps um known from from gitlab ci cd, where you just say hey, I don't want to test it in production right now. The button changes from green to yellow. I have a deployment in kubernetes, for example, in a container, the same should be possible with a separate ec2 instance, where you just see what what is in the preview. What is happening in the merge request in requested branch, long story short. There is more more ideas beyond that, but this is an ongoing project of ours.

A

The template is, should be soon merged to to gitlab core, so maybe it will hit 13.7 or maybe I think, yeah the freeze window is on friday um or maybe one conversion later on. The cool thing is um since the deploy template project is a separate project, a public project.

A

Everyone can just use it or try it out now. I wouldn't recommend putting it in production currently, but you can try and.

A

Just moving the zoom a little bit and I have too many tabs opened again.

A

This is basically the deploy template. um It's a repository. It contains everything which is needed um just to you can peek into the uh the thing, and I will link it in the in the video description later on um the main ideas of them to provide you with the things to get started, everything which is provisioned what is needed. So basically, you need to navigate into your project settings in the ci cd section.

A

Add those variables mask them so that they're not like exposed in the in the droplog log output, and then you create a new gitlab crm file and just copy paste. This and add a real email address for the let's encrypt certificate.

A

That's about it. The rest is more or less um happening automatically, and you can configure many things um like, for example.

A

What would we we could configure that port? So if the the docker container or the the application exposes a different port than three three thousand, for example, 80 or 8080, or something else, you could just go with that and modify that from within the configuration.

A

Long story short. um Let me see we so I had that opened in the german caffeine two hours ago.

A

I started to learn vu chairs uh today and I didn't I was really using it. I I know how it works kind of, um and I have seen many videos, and I also know that we're using it for for the gitlab product, um but I didn't really start with it, and so I kind of looks for a tutorial an example uh using the cli command and generated an example application, and then I decided, okay, I might modify it a little bit so adding a new header and, um in the low end, changing the logo.

A

um And last but not least, I wanted to create a docker image, and this is basically um just to show you that this is for real. um I really like that, um especially um it's everything is documented and you can like just go ahead and copy paste it um and use that so like it's building um the entire application with npm.

A

To be honest, I don't really care how it's built. It should create an image which can be run by the five minute production app workflow, and I can just like focus on on actual development.

A

The thing is what I learned with with wu. You can run your local dev environment with yarn. I think it's yarn yarn surf and you just see what's what's going on um and for the deployment we just needed a docker file, so I basically copy pasted this into.

A

A new docker file I needed to fix some things and learn some things on the way, but this is basically it and the second part I did next to the application um was just to add everything which is needed for the deployment.

A

um I've modified the application port because fiddling with nginx and docker and different ports was didn't work, at least not in the five minutes I got but yeah. This is basically how it looks like and for the source code for the application itself.

A

It's just the structure which is grounded, um and then you or then I started editing the hello world section and philip mentioned earlier today that the components hello world file got the rest of it and we can like dive a little deeper and we added a link to our everyone can contribute dot com site already, and the cool thing is, for instance,.

A

And I want to trigger a deployment right now um we could edit something. Maybe what could we add here? um We could, for example, link the blog post.

A

A

I could totally do that in a different uh terminal uh on a local terminal with git port, but right now just too easy. uh This is p and this is slash p self self html for the win.

A

And we have our.

A

Href loop target blank.

A

How is it called first block? First.

A

Deployments five minutes, question mark and there's a spoiler inside it's without kubernetes, um so the whole idea is to make it as easy and approachable as possible and if you need to debug something um you can just ssh into the ec2 instance, grab the locks or do something else around it with the ec2 console um and you don't have to care, you totally can use auto levels with kubernetes, but right now you don't have to care about anything which is related to kubernetes. um We have added this. Does this make sense?

A

Yes, we will be adding this and just to show you how this works. I will skip the merchandise part and commit it directly to the main branch at blog post insights.

A

Because this triggers the pipeline, which has three jobs and stages, the thing which basically happens here, is building the docker image.

A

um This takes takes a while, um so I might just be hopping over to the introduction website, um which has been deployed already so it's it looks like this.

A

Yeah, this was the thing we added before and we want to add some more things onto the website and work on that inside the ec2 console one can see all the details and this instance should be running since.

A

Launch time: oh that's the different instance.

A

This is the newest version.

A

It takes a bit longer. Okay, it's still running so I I already started that and the next deployment is just going on so the first provisioning takes takes a bit because the server must be provisioned and also the database host. If this is a to-do of mine like adding a database backend for this application, but still- and um in this in the future, like the two to n iterations, it's faster because terraform sees that there is um the resource already no need to provision it again.

A

It's just redeploying the application actually and restarting the nginx front end. um So we do have here is the pipeline still running.

A

It's not super fast. Today. um Npm is okay yeah, so building the the image takes a bit, um but still.

A

And you can see I've.

B

A

It worked all the time interesting.

A

Do you have any questions thus far, or did I miss anything in my explanations.

C

I have a little question, uh so the you have your repository, your five minute app and then there is a gitlab ci template file, which is another repository which I have to include that I can do with right.

A

Yep yep and the future idea is that the template is provided out of the box from gitlab itself so similar as you would be, including, for example, sas scanning or code quality scans.

A

We provide several built-in templates. um Just to this is uh it's it's, the cia directory. um Do we have that over here? Yes,.

A

Is this what no? This is not what we provide. This is our own test. Sorry, um I need to google.gitlab see our templates.

A

This is the path actually um so when you, for example, um choose to use the z plus plus template or the docker template from within when you create a gitlab crm from the editor, it's using the content in there, but you could actually say hey. I don't want to like copy paste. What's what's in there, you can just reference these examples or these templates.

A

So, for example, autodevops user uses that method, but it is also super easy to say hey. um We have, for example, for security.

A

uh We want to add fast or secret scanning or secret detection to our pipeline, and this works in a way of saying, hey, include and then you're listing. uh What's what's going to happen, and I think we have it for the ci community day so is.

C

It possible to combine multiple templates so once us.

A

Let me see if I can dig it up from the ci secure. Let's ah yeah, this is the one I'm looking for. So I did a security webcast a while ago and one of the things or one of the ideas is um you go there step by step or by iteration. You start with.

A

I want to do secret scanning, for example, then you add fast so static application, security, testing for detecting chmod or file operations and things which can be detected from static code scanning, and then you can combine it, and it's also so templates are referencing. The built-in templates.

A

Let me just open up the documentation um include template, no, not docker. Today's today is not my day. um There are different ways how you can include using clues, and I think this was one of the questions yesterday in the in the mai as well. um The most simplest iteration is just you have one template, you have the string. um Multiple templates can be included as an array notation or list notation.

A

um You have the possibility to define remote templates. So this is in a different repository, then, must be public or at least accessible, um and this is currently what we're using as a workaround for the grandson's sunset as a as a possibility.

A

While we are developing a template, um you could you can still reference it as as a remote template um and over here you, you see a mix of I want to like. Did it before script, template, which is a generic one and say hey, I want to always export some variables or install something or check something in the before script. Section later on you navigate into having a local template inside the repository which defines after script.

A

um Still, the execution order is not like like this here, it's just what body is defined and it's compiling that this makes it harder to debug when something is not working.

A

But as dove mentioned yesterday, there are plans to provide you with the full compiled output um similar to a country compiler, basically, that you can see what's what's actually being executed in the ci cd pipeline.

C

Does it does it use the the so? Does it just copy paste the the yaml file template file, or is there any special requirement for the template so like having a special name and use it um stuff on it or.

A

So basically, um this template notation needs to be either provided by us or by the by gitlab itself, or you could say um I think.

A

Our documentation is huge. um Google is fast gitlab ci project templates.

A

um We do have the possibility to add um ci templates.

A

As far as I know, not description templates.

A

And I think that's an enterprise feature to add your own.

A

I'm just not sure about it right now, um but just to to answer your question, it's the um the format is the same as you would write your gitlab cr yammer file, and probably I should have opened up the not this one, but too many tops this one.

A

If you look inside, for example, for sast, you see lots of things, but in the end you have the variable sections. You have the job definition, which is fast. You have an inside template which can be used which can be included or imported and then that's basically chop definitions, and this is heavily using rules.

A

So if you want to get an inspiration for for ci rules, this could be an an interesting starting point, especially because.

A

It's checking whether specific files exist and only starts this scanner. Then, for example, this is also a reference to the um rules and changes, detecting a change um and then triggering a pipeline or not. um So this is um basically everything which is best practice is hidden here and you don't need, don't really need to care about it, but you can use it as an inspiration for your own um templates.

C

So I would guess: wait if I have two templates, which has a name collision, maybe same job name or something like that. I would guess the ci system one uh will throw an error that there is a collision or something like that or.

B

C

Late later on, wins, or something like that.

A

I think there should be an error being thrown, but, to be honest, I haven't tried that um I was so when I started with with the language. I didn't understand that the jobs are free form on on the left on the first level. First and my brain automatically decided to oh I'm trying to keep this like unique as a name, but I think.

A

You cannot define a job twice, I think in my in my theory, there should be an error being thrown, but we can test it.

C

Yeah, if you have two templates, uh which has the same chop name for example, because I know systems which just add the prefix from the includes to avoid such things.

D

Okay, okay, this could be probably depends on the merge, so I know in one ci yammer, you can't have the same two drop names.

A

No, we want to include something we want to include.

A

Oh, that's nice.

A

Template file: okay, what's the difference between finance, okay, um let's just use local and we say local repository, the repository.

C

To the repository.

A

A

Let's just try it. um I need two separate yammers.

A

A

And I want to create yeah merc 2.

A

And this is the magic now. This is the one it doesn't work, but it's a strange bug. um I want to include this file and I want to include this file.

A

This is the setup and over here we just say job one, and I think we need to at least have a script section which doesn't equal, zero and exit zero and that's about it, and this is the second part. I think this should be the simplest example to make it break. um Should we end.

C

In an echo to know, if the last one wins that we know which one wins or.

A

I'm I'm thinking that the compiler will break so we'll we will get the yamaha error right now. Okay, this is, I think, yeah, it's a good thinking. uh Can I just cut that.

C

But I also guess it it's just copy pasting and then reading the cable file so.

A

A

Good lab, I don't know if you can get the file name, I have no. I don't think that that's possible.

C

A

Yeah I wanted to I wanted to print the location where this is executed from, and I've developed, uh dsls myself, where this was possible, um but I bet it's not possible from within the yammer configuration language. It's a mix of dsl and configuration language.

A

So, oh yeah, that won't work because I'm and the copy paste hero today um this one and we say.

A

Templates, no, not templates, includes same job name.

A

Test excludes this.

A

A

Oh that's interesting.

C

A

This was kind of that's not expected, um but it's interesting to learn. Learning learning by doing.

C

A

Yeah, I know that from was it mario's. I think so, and this was kind of unpredictable because you had 10 different inclusions.

C

So it's it's easy, so you can it's! It's like inheritance.

A

ah Okay, it's using a template, template inheritance, so last template includes wins. It's not actually like the the files on its own, because if you write it in the same file, I think we will get an error yeah.

C

So it's not just copy pasting together, amplifiers and then read.

D

It, but I think it's already also documented in the documentation that they say when you, when you're drawing merge something that the last one will win, but I don't find it right now.

A

Yeah- but this is like in this is an import template strategy um if, if we would go ahead and copy paste, the content into the single gamma file, which I'm trying to do now, but um this shouldn't work- and this should definitely because we're in the same scope, we should get a compiler, config, compiler error.

A

um Let's just go ahead this one um and we're using git, so it doesn't matter.

B

A

Write it, we can just revert it later on.

C

Maybe this is a feature suggestion, but it would be cool because you already mentioned that you tried to build something that you can read the text. Maybe it's cool, maybe it's cool to resolve it.

A

It's checking the keys and it does that live which um prevents me from committing this fool and testing that um with the with the linting later on, which is like it's not efficient, and I'm glad that we are solving this. um It's nice to create a screenshot from that.

B

But does it actually prevent you from committing it or is it just showing the warning.

C

I guess it's just showing it.

A

What you want, you think it is like have.

B

A

I think I think it's a good thought that maybe it could prevent you from committing it, but maybe you.

D

Want to find it intentionally, yeah yeah.

A

D

Case are pre-chromators are mostly, and you can do it locally on your machine or on the server. So you could get also.

C

um Access to the local instance to do this right.

D

Yeah to do server hardware yeah, of course you can't use it on versus so on, declub.com.

D

But okay, when you use git port, you could have.

C

D

And then it doesn't will appear.

D

A

Found it every time I put.

D

Away so probably, if someone I used pre uh chromatograph right now, since two months I didn't use it before, but right now it helps me a lot to see.

A

A

B

Not it makes sense actually.

A

No, it makes sense actually because it's using the same logic as with the imports, so you can override it, but still it's using the same key. So the ui is is more intelligent.

D

A

Telling you that you created something which overrides it but from like definition, logic, if you define it later on with the same key, it overrides it.

B

C

B

But aren't using.

C

The keyboard, it's just the keyboard structure, so it's just overwriting, that's it.

A

It's a dictionary, basically, where you add the same key after the thousand iterations or a thousand keys. You will write it.

B

C

Be interesting.

B

um Importing both files, I think we.

A

Cannot mention the first file.

B

And we still import the first and second one right. I'm.

A

Not importing any file currently, um this is like the plain uh the full ci template, so the current change is these files are just sitting around um it's this one and.

B

A

Duplicate key makes sense. It's a warning.

A

But it could be the case that you definitely want to.

B

Overwrite this maybe.

A

This has something to do with extends and using skeletons with the same name. Maybe I don't know it's weird everybody's. I think it's good to to have it right out and know about it.

C

A

C

They're they just used plain yellow. So it's.

D

D

I think they use some public uh ruby gem, but let's, let's look into the code so.

A

It's uh it's a yummy parser, but it's um the thing is for, um if you want to provide an error, mesh message, for example, for duplicate job name, um you need to define own rules and also like how to say that um you, you read basically the um the objects into into an object, tree and structure.

A

So there is no logic yet and later on. You define, oh I'm in the scope now and for example, um that you can throw a warning through an error when a script is missing.

A

This is required for a job. um This is a validator which throws an exception and provides you with an error message which you can read, um because otherwise it would be just some like in this file in this line. Something else happen. This is also which we see some time um and then you you do sort of like look ahead and say: hey, I'm now in the job scope, when the job key was scoped, then I'm within script, and if I'm, for example, you missing the dash and a specific error thrown um in this case.

A

It's easy because script um script is expected to be an array or a list. All the time, so, if the dash is, is missing, it's a generic array error.

A

The problem is, if, if you don't really know yaml as a user, I have no idea what it tries to tell me and one one thing I'm always failing with is the syntax for the various, and this is this kind of brings me back to this one um or the the five minute production app actually which should be finished by now, um and I just need to find the red top only no, I I changed did I changed.

A

This was the red top, um but actually I want to have a repository.

A

And yeah, in my my brain says: if I have a list, I want to have dashes, but variables is actually a key value definition and the first error I'm usually making and I'm getting better at it- is removing the dashes from the list in the next commit.

A

So I'm also looking forward to the ui pipeline editor, which just gives me hey: hey value, yeah have fun um basically and and more things like this.

A

um Okay, the pipeline has been run so adding our changes, and I think nicholas, didn't see that then, before um we kind of wanted to modify- and I think I didn't refresh the page yet um yeah- we wanted to modify the five-minute production app.

A

I tried uh today at the aws reinvent, live demo um and learning vue from scratch, and we wanted to add something to that and in order to do that getting back here um we have, we see the pipelines and I, for example, want to see the permission step stage, which is using terraform plan and apply.

A

I don't speak terraform, it does something in the end. What it does is provision the t2 micro instance, if not existing already firing wrapper database ec2 instance, if not existing already, and it also manages tls and other things, and one thing it also does. Is it populates the environment and provides us with details on uh on the provisioning side, because we need to appear address and and other things this is all hidden inside the cr template right now, so, as a user, I don't care about it. um I just want to have it deployed.

A

The next step is um to use or to go into the deploy stage, and this is basically similar to to the rest it. um It builds up the instance with an nginx up front for the web. Application sets up all the proxy stuff pulls the docker image which has been pushed to the gitlab registry before and.

A

Kicks off less encrypt, this is why you need a certificate email variable just for let's encrypt. This has been merged. I think two days ago, so you get tls or https for free and the other part is. It generates a nib io domain to avoid an ip address, but you could also define your own domain in the end, which I will be doing later on and yeah that's about it, and if we refresh the website now, we should see.

A

Our change- and this is now live in production. um The thing which we currently don't have, I think, is you the ideas using the revenue apps in the same fashion as with kubernetes um that we fire up and you a secondary ec2 instance with that specific branch. And then you see that the tanuki rainbow change or the the page update how it looks like and when the revenue goes.

A

Okay, you approve it, you merge, it gets deployed and it's ready in production um yeah, and I just learned who today in five or ten minutes, it's really nice. I need to dig further into that and I probably if I can learn it in five minutes. You can learn it in two minutes um now it's it's really easy to get going with a local dev environment and also it's nice to see to just have it deployed.

A

um The thing is.

A

What's that a class learn x.

D

You can learn every language in x or y.

A

X x, to the c exhibit.

D

So now you make it for our pipelines.

A

Technically, uh you just need a container which exposes the port. The problem is, if it doesn't speak, http it's. You won't see anything currently.

A

So one thing we we decided on the way and this somewhere we have the uh the blog post from today. ah By the way I could show you the oh. This is something else: where is the ec2 console over here? um So it's still running it's using the so I started it five hours ago for the first time, um it's still running it's using the uh that's the t2 micro thing.

A

The user interface is not not made for me, um but yeah, it's not generating any cost right. Now, it's the the freight here and yeah. If you want to build your own website or maybe just like not use anything which is pages um just go ahead and do that, um what else did I? I was looking for the blog post. Actually, where is the blog post.

A

This is this one.

A

And yeah future plans are to merge this upstream, but you can still like test it out right now, as I did and follow everything which is like the direction there is the issue board and currently watches in doing email, service.

A

Yeah and and other things.

A

Indeed, I think I missed something.

A

Yeah, oh ah and the the providing the aws uh credentials in variables done in the project settings. um This should be like moved away and saying: hey you're logging into gitlab you're logging into aws, and then you get a list of your eim profile of what you want to use for it. You just click it, and then everything is done. So um this is basically the the automated way we are currently using or the most easiest, iteration and approach um but yeah.

A

What else can we hack or do do we want to try something else.

B

A

But have a question.

B

um Regarding the templates we included, can you override the variables by including the templates like in the exhaust template? There were some variables defined on the top and it doesn't use the environment variables there. Can you overwrite them.

A

You can actually override specific variables defined in in existing jobs, and this is how you configure the. Let me just again show my screen.

A

I need to find the top. um Where is this one? um So you can see and where? Let's let us ignore these, um these fast and secret scanning, they are defining specific variables inside the jobs and one of one strategy can be to say: hey, maybe I want to use. This is a good example.

A

um The secret detection has the has the the historic scan disabled by default, because if, if you got 10 000 uh commits it would be scanning, 10 000 commits- and this doesn't scale.

A

And you can overwrite this variable and then say hey. This is true, because otherwise the default would be false and by you need by. uh In order to do so, you need to pick the same job name and then override, for example, the stage the variables and also, for example, the artifacts.

A

This is I copied that somewhere, um this is not needed. Actually it's just. If you want to overwrite a variable and say hey, I want to enable something or want to change a path or even change the stage which I did for some reason.

A

You can do so or enable the debug trace, because I wanted to see what the scanner is actually doing for sas, and I think I was using a python project yeah. I was actually trying to understand what the scanner does it's running in a docker container and it does everything automatically, um but I really wanted to see what it what is going on, and this is possible.

A

um To be honest, I don't know where the generic documentation for that is.

C

um Now it makes totally sense why the last one wins.

A

Yeah, because you can actually overwrite it's, it's add, add more key, add more attributes to the object, um but also overwrite the attributes. This is there and.

C

Is the case that when you have the, when you use the extend keyword, you can use the name without the dot right.

C

So when you use the extend key for a job, I guess you don't have to specify a dot in the.

A

uh Too many jumping um it's.

C

It's it's not really.

A

No it's um when it extends which looks interesting. What do we have here include.

A

Extends sonar analysis.

A

Okay, this would be a different way to do it.

A

It stands for that.

A

Was really hard to learn? I learned it from a friend of mine who used it and was like looks interesting. What is that extends is something from yammer.

A

Oh, this was the blog post. I was actually looking for. um Twitter is my personal bookmark list.

A

This is the yaml thing, keep it simple and yeah this is extends so using a dot, basically hides this as a template, uh if you, if you so, this is not instantiated as a job object.

A

Does that make sense in in other languages you would be using or where I'm coming from template as a as a specifier up front and then object and objects get instantiated and templates are not. They are just using as a reference um and test extends this. So basically it inherits the image, but you could also override this again saying hey. I want to use a different image for this job.

A

Let me see so I'm putting this in my in the zoom.

A

I think maybe I need to look it up somewhere. In my this is a different history. You cannot see right now, um yeah this one.

A

So marcos is a good friend of mine and he's really like digging into.

A

um Ei configuration optimizations and I think there are plenty of.

A

Things which use extends and also.

A

This way of import, so I think this was this- is a a built-in way of not so not so beautiful or not so easy, understandable way of using extents. So, basically, this extends the test template which is defined in.

A

I think there yeah.

A

Yeah, it's actually it's using a different synthetic and in the in the beginning, it was really interesting to see that you can refactor your ci configuration code um into using templates and generic definitions.

A

But in the end it's like it's uh it's a dance on the volcano either you have it easy to understand or you have it super organized and and everything else, one thing which, which is a good strategy, especially if you're creating the infrastructure for your teams.

A

You are providing default, ci templates, which set specific variables which define what's what's going to happen and your developers can just consume it or maybe they they are not even allowed to change it um and to define their jobs, and everything else has been. Pre-Filled has been preset similar to the, uh for example, for the five minute production app template. This gets included, and if you want to define your own chops for ci cd, for unit testing, for example, or for building the building a docker image with a different scope, you can totally do that.

A

It's just reusing, what's already there and avoiding duplication, but also making sure that no one can modify the central ci configuration, for example,.

C

Okay, so is it possible with x10, maybe because I know that they're interesting in general, you can't add values to such uh sub childs. Is this possible with extent? So if you have a.

B

C

Chop with labor or artifact use labels or text, and you have two in the in the template and then you want just add one extra value in the extended job this I know that interest didn't work not in the way I was thinking of.

A

Can you probably be something I can google some some some keywords to find an example.

C

um It's it isn't that hard, because I know the workaround but um think of it that you have this. Is it called labels for text? I think it's called text in cio. It's not labels. When you can specify your runner, uh which is yeah. I think it's text. It is an example and you have.

C

For example, I have a windows tag and I have a windows, 10 tech and I want to inverted the template and I just want to add one of uh of this text, for example- and I know that with the entrance it only works when you have inserted directly under the under the level. So I have to split it up into multiple templates.

C

It would be nice to know if the extent mechanism- maybe could just do this out of the box, because this this this hits me hard at the beginning.

A

So you want to modify the array I want to just extend it.

C

Or I think remove is hard that way, but just extend it because I only know that I have to write some anchor point at the top and then insert it directly under the text. But I have to use multiple of these entry points, which yeah is not that cool.

A

I think that's the limitation of yammer yeah to be honest if it would be.

D

Probably do a workaround with uh the new metrics drops.

D

B

D

It's not new. It's I like two releases. I draw all three releases or something like that.

A

What do you mean metrics.

D

Where you can bid in peril and specify which text should be used for varanas.

A

ah Matrix I've I've, I understood metrics, which means monitoring.

D

Metrics no metrics like the metrics of neo so but for ci.

A

Yeah, the meme when someone is says that ci is in cd or something.

A

Interesting, um the ci metrics, I'm not sure if you can solve it with that.

C

But the interesting question is: maybe if there are, I think there are some differences besides the fact that it's nice to read this extend key so.

A

I think you will carry on with your work around, um but it's a good way. I don't know your work around to write, write it down somewhere um hint blog, post hint um or maybe just share it somewhere and we can like take it. Take a deeper look uh next time.

C

Yeah, it's it's not that hard, but I like the extends keyword much more than the the young launcher points in github chat.

A

Yeah, maybe we can also collaborate for maybe for guest blog post on our blog or something around this, because I think it's a hot topic- and I know that um chris is currently working on some insights, and this was one of the first ones. Oh there's a this shouldn't happen, something is broken um xml, that's the problem with html rendering and the.

B

A

A

I need to fix it later, um but I think.

C

He uses this: can this a little bit rare way because he uses this the job? Definitely, okay, it's now instantly.

C

You know, because my background is just make a second dot template and just add the star job definition thing underneath this text thing so then you can insert it into the point.

A

Yeah, I think yeah manipulation of of existing storage is or existing attributes is a challenge, and I think it's probably not solvable in yammer, and I I know different languages and dsls, but this isn't possible either and if you build it in a way, it's opinionated because it uses what python does and then someone else comes there, but in php it works this way. Can we change anything? No because in production it works right now and if I change it, everything breaks- and this is, I think, one of the problems also with problems.

A

The gitlab ci yaml configuration language has evolved over time and if you want to like remove something or change the behavior like duplicated objects and shop names, if you would like throw an error now, I'm not sure how many people uh will have a broken pipeline, because just because of the upgrade.

C

You will get a lot of friends.

A

I think that the main issue is the one who does. The update is not does normally not fix, like the ci yammer thing, so it involves the team effort and and causes headaches and probably will never change that behavior or maybe in a version three or version whatever is coming in the future, not version three. We could break it and change it, for example in 14.0, which is coming on in may on the 22nd. um But on the other side, is it worth the effort which which use case of which, which problem would be solved?

A

Oh, it's an error yeah. Maybe we can build better, tooling or better visualization of of the error and also like pre-commit hooks or something else which helps solve the case. If we cannot touch the foundation or the language itself, this is at least how I would do it because in my past I've broken many things and community members were not so happy about this and I totally understand it.

A

But if you're young and adventurous, you probably don't know how much, how much effort it is to fix configuration on many servers deployed in production and you don't have anything like ansible or puppet. Yet.

A

A

I hope this was usable or interesting anything else which comes to mind, or should we say goodbye on youtube.

C

I want just to say it's cool to say a five minute production app to deploy that easy, because for many people this could be really helpful.

B

Yeah absolutely- and I think this might even be more helpful for running tests than just deploying something, so this could be really nice.

A

That's a good point: I'm actually like having the the monitoring and observability portion in mind where say hey. I want to know this git commit decrease the performance or the the memory usage or something else which is a base metric by 25, and I want to see that integrated in my merch request. Right now I can make use of of kubernetes and the prometheus application as a sidecar and the same idea goes with well.

A

If we run the web application, we can also, for example, spin up a promisius node exporter instance in docker, and have that available and scrape that somehow. But this is like the 10th iteration beyond that, but it would be really nice to have that and see hey.

A

If I add this amazing function, which has a thousand node modules dependencies, um the code gets super easy to read, but it's 10 times slower and my users or our users are not like getting something out of it.

A

So yeah look looking forward to what's what's going to happen, and especially um if you want to try it out and um like do it, do it in the future. Just ping me anytime, and we can do a spontaneous session um or just um also for for feedback, do not hesitate um to ping us and create issues and or maybe just try it out in your preferred language and write a blog post about. It would also be super nice.

A

This is not. This is not only meant for you, but also everyone who's, listening uh or watching on youtube. um So the the more we test, the more we try it out, the better it gets over time.

A

um Okay, then, thanks thanks for listening thanks for watching and we will see each other in january, so happy holidays and.