GitLab Developer Evangelism Team, 16 Sep 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Develop CI/CD best practices with GitLab - Workshop at Collision from Home by Michael Friedrich

Description

Development is more than just code, and last-minute bug fixes on release day can increase stress and lead to burnout. Guided by real-world exercises, learn how CI/CD improves your development workflow, and practice automated build-and-deploy stages.

Slides: https://docs.google.com/presentation/d/1j5ROpjwmBiXqtiHN6lJ20UcJCefw4-KC25lWWoDqTnY/edit?usp=sharing

Repository: https://gitlab.com/dnsmichi/collision-from-home-2020

A

Welcome everyone today, I want to talk about developing the ICD best practices with kidnap, and we do that in a little workshop today in the next 20 minutes, I'm going to provide you with a short introduction mine over here, the first steps we are getting started with a hands on the ICD project, so which means we are using a team or go project to test, build and run it and look look after CICE. What's next after the workshop, to give you a little introduction about myself, Michael I'm, a developer evangelist at gitlab and I.

A

Do love things open-source so before joining kit lab in March, I've been working in the zoo's monitoring project area and when I'm not affiliated with anything in open-source I love building Lego models, so the Lego hugglebots model will be my next passion.

A

If you want to reach out on social media and then in a way else, I go with the nsv key, which is the lovely version of Marty in German. And if you want to learn more about technically evangelism, please be sure to check out the kit lab handbook. But that's now that not about me it's about CICE! So, let's get started.

A

Why would I want to use the ICD continuous integration, continuous deployment delivery, depending on the situation? So when you keep working on a specific new feature, you want to collaborate with your colleagues, your teammates, but you don't want to disturb them with your changes because they might be breaking. You want to try things out and come iterate on specific features, algorithms and just keep working in your own space.

A

So that's the reason why you can create a new git branch and you just keep pushing your changes and that after some point you might want to have the review and everything is merge back to master, which is the production after taking any release, and often the release is just okay and because you kind of tested it on your laundry machine, everything looked fine, but the production uses a different Linux distribution or something else.

A

And basically, you need to need to work over hours or after our fixture, released and fix the next release, and everything else is just not so not I'm using about development, and you want to make your life pretty much more easier, more relaxed and also to you want to ensure that everything runs fine. And this is the point where you might have, for example, written some unit tests or some end-to-end tests for your application already, but every one is random manually and with in github, see ICD, you can automate these steps.

A

You can define jobs in your pipeline which automatically run the tests you created before, or you might need to add them. You can also deploy your code or your website to review application, which then highlights exactly the state. You have just committed and changed for someone to review, and you can also like test specific things.

A

One thing which is also worth mentioning is that you can also integrate security checks in CI, CD, meaning to say application, security testing dependency scanning so that nothing gets deployed to production which might contain a security vulnerability and that's basically, the entire workflow and it's overwhelming I know. But we need to kind of find a way to start simple, and this is something which we are going to do now. I've prepared a project over at market net namespace. So you can just check this project out. You can fork it and try it out.

A

Once we go along or maybe you will, you watch the video afterwards and you can just learn how to use gdb ICD. So the thing is when I navigate into the project, we will be seeing a little coding project with the Maine dot girl and some library code and way enough. For example, inspect the Maine dot girl I can see. Okay, we have redefining a main package and we're calling the main function, which kind of prints code and we're getting like read it.

A

Hopefully, it calls a function, but we just don't know what's going to happen and basically does something, but when we're not really sure what it does, we just received the code from an external vendor and we got the exercise to add our own features to test it, to build it and to ensure that everything is running smooth.

A

So the first thing when receiving for code from someone else or when I even want to start the project is I'm going to think about my built architecture, my birth structure and how to add automated testing and automated deployment afterwards. So the first first thing I want to do now is I want to black say to my colleagues: hey I've received this code now I'm working on the project and I'm, going to add a new task for myself.

A

In order to do that, I'm navigating into the issue moves and creating you my store now so I want to create a new milestone, say: ok, this is the version 1.0 and that you date is our workshop session and I could say in the description. This is my CI CD configuration milestone.

A

It's just create a milestone, and the next point is next step is to create a new issue which says: ok, we want to add CI CD configuration for project, and we also want to use one of get labs, hidden powers with task lists, so I want to add the configuration.

A

Of course I want to verify the test, stop being run and someone needs to read some dogs, so my colleagues know what's going to happen in order to show you, we can have the check boxes and once we are going to create this issue, we'll assign it to the milestone. Now we can see that we can iterate Li iteratively work on under specific tasks and take them if what, if needed- and it's also highlighted over here and also in the issue listing so everyone knows what the state of the issue is.

A

Now that we've said we want to create a CI CD configuration we actually to do it, which is the next step, and in order to achieve that, we don't need to change into any shell or something like that. We're just staying in our repository and we're making use of the web IDE, which is a little bit hidden over here.

A

So this is a browser, integrated interface, where we can just edit the existing code, but we could also like add a new file and we want to use this specific target left, see I, don't Yama file, which provides which defines the CI configuration in this specific example, I'm not going to write some some Yammer code now because it's really time consuming, but we do know from the documentation that there was the ISEE configuration template files and luckily, for us we do have one for Galang.

A

We can just select it if it worked. Yeah it worked, and then we get basically a predefined template. I will now just edit some parts and explain them to you later on. I just need to ensure that the repository name is the same collision from home 2020 and just checking in move on took later on work under tanuki.

A

Behind your way, I'll explain the steps just in a second I just want to ensure that the CI CD configuration is being run in the background for now I'm now creating a new branch I'm committing that I'm, saying: okay, I'm, adding the CIA city convict now I'm, just creating a new branch and College feature: CI city config, and also creating a new merge request.

A

The cool thing is when I've just added these changes created a git branch which would be an manual step in for now. I didn't leave the context so I'm still within get lab. I can create, create a moat request. I can say: I want to use fixes and then the CI OCD configuration for project. So when this merge request gets merged back to master, we want to auto close the issue.

A

I can assign it to myself, might assign a milestone, yeah and then, let's just submit it and see if our source code actually compiles and what's going on. As you can see now, the Parkland is running the pipeline is, is running on get lapto come within the shared runners, so I'm sharing my resources with other users. If I do want to use my own runners, it's totally fine.

A

So in this case the job failed and I can see within my pipeline's. Okay, it didn't come. It didn't continue in that regard, so the former job failed and the compiled job did nothing. I told it before I wanted to show you what specifically is going on so we're changing back into our merch request and I will show you within the web ID, because we can edit it the partland failed and now were just looking what's going on inside.

A

So the thing which happens here is it pulls the goal-line docker image prepares something for building the the Golan environment and then it defines the stitches. So first off there was a four-month chop which runs in the test stage. Then there's a compile job in the build stage.

A

Whenever the test stage contain completes or continues, the build stage will be fired in our case. Go vet, for example, broke, and the thing about here is when I change back to my CI CD output, I can just see. Okay format has has failed, doesn't know about the specific packaging killed. So the thing is I kind of figured, maybe the import 44. The golden code is missing. The next thing is to navigate back into our merge request and within the mush request. There's the opening lap IDE button, where I can just fix the code.

A

So previously we've been in our CI configuration, but we can also write : code inside over here and we can fix it. So if one of the fixes is to just reuse what I've commented out here, just the what ran in too much code and with importing that it should fix our code, I want to commit inside the branch and fix missing package import edge to see if this works committed and the cool thing is once I've committed this to the branch Oh to the merge request.

A

The CI pipeline automatically starts in the background and with the rocket icon over here. I can just check this out and see. Okay, it started building, and while it built I, can also create it in a new tab over here and just see what what else is going on whenever four month will continue. It also runs tests and the next top stage, with inbuilt, will actually compile the government binary. So it's running some some arguments for Coke built and it also defines an artifact.

A

So the binary, which is which is output, is put out from Co build, will be available from kit lab itself, so the runner uploads the artifacts in this case the Tanooki binary and hopefully the truck continues, which it does. Let's just go back and check over into the pipeline, which means when the first shop continued. The second one is actually compiling and again I need to wait a little bit because every job is just executed in its own sandbox.

A

So that's also one of the things you always get a reliable test environment, so every job which is being run is basically the same, and you don't have any like changes from from a previous round which installed a dependency which is now there. But it's not it's not expected to be there. This is one of the big advantages of using container sandbox and always using the same environment. As you can see.

A

In the output hours, our script checks are being run and the binary has been built and to run uploads the artifacts to the github server, which then means okay, I can just check out the top artifacts over here and see. Oh there's, a don't be binary. I could download that now and just try to execute it, but it's kind of cumbersome and I. Don't want to do that. So I can actually verify that this is being run, and this is an this is.

A

This will be a chief by adding another job to our stages, which we will be calling run, and we will like similarly simulate a deployment now and run the script Tanooki.

A

In order to do that, we're just navigating back over here in our CI CD configuration and we can see we have the deploy, stages or stage already so. In this example, we are just moving at the bottom and say we want to define the run job. The names are freely definable, we're adding a tab over here, adding it to the stage deploy, can also use autocomplete.

A

We are using the script keyboard to execute something top and execute the Tanooki panel, and that's basically the thing. The third job being added and we commit our change and say add, run, show and actually let that run and we can see o on the left. On the right hand, side, the CI pipeline has been restarted and it runs again. It doesn't cache. The previous runs because, right now we have committed something. So we changed the source code or the CI configuration.

A

So it's entirely started fresh and the pipeline's keep running, and while the pipeline keeps running as we figured that we're pulling the going image and it kind of takes a while, we can just check out what else might be needed of what else might be useful.

A

So when, when you start coding, you might not just use goal, and you might just use a different framework and golang has the test framework, for example built in but Python or C++ have different frameworks, so make sure to check them out and to read scripts or item to your CI configuration within our garnet project. This has been achieved within the Tanooki underscore test that goal so basically that Saluki dot goal implements as a function which provides a string.

A

Let's call it like that, and the test function just wraps that with within the test framework and checks, whether the fire to return specific string items- and you can just like build that for your own functions and extend it and it's automatically built-in when you run, go tests and get your results back then, and this is a pretty cool thing- our trips are still running and the next thing is for self-practice.

A

When you open up the EMR when it completes, you might want to also close the issues with edit the fixes hash one before you can alter close an issue. So when everyone approves the merge request, it gets merge back to master and the issue gets auto close, so you'd only have to use an extra step to automatically close.

A

The issue can also specify to the branch gets deleted once it merged through your repository and your get lab project is clean as it was before, once everything has been merged back to to keep mouse and the issues have been closed out for the milestone, creating new release, taking new release and also links to milestones or everything is combined.

A

The things which happens which happen next for your CICE pipelines, should be using environments for diplomas, we're using staging and production. You can use rev your apps. You can also think about deployments with a shriek of Terror for more read at the end support just using the get-ups workflows. You know you should also think about adding specific monitoring to your CL CD player five lines. So it's not just about the job fate, but actually your application might deploy to a kubernetes cluster or something else. So you so.

A

Each gate commits by the influence the performance, the application performance or the memory usage just keep monitoring that and you get a pretty pretty pretty exciting possibility to analyze and report, while the performance was degraded by a specific git commit which kind of changed the backend code and the nurger with and last but not least, I want to remind you, but security like runtime security and also static code analysis and allows his dependency scanning so why you are not responsible for word and last but not least, werner ability management, let's just check whether they deployment was okay and we can see.

A

We've got our fancy tanooki and everything worked fine, so right now we would be good to go too much too much request, but that's just for you and with that thanks for for listening. If you got any questions, I want to learn more about it. I've linked to get lab training, but with that I say thanks and waiting for the questions.