GitLab Developer Evangelism Team, 1 Feb 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 1. Switzerland GitLab meetup: First time GitLab & CI/CD workshop with Michael Friedrich

Description

- Repository with exercises & slides: https://gitlab.com/gitlab-de/workshops/swiss-meetup-2021-jan
- Agenda with timestamps:

* Introduction 4:32
* First steps with GitLab: Labels, Issue templates, etc. 9:12
* CI/CD: First Steps 18:13
* CI/CD: Go Project & Tests 33:07
* CI/CD Pipeline Editor 46:45
* CI Security: Detect Secrets and Vulnerabilities 51:03
* What's next? 59:00
* Exercises for Async Practice 1:00:57
* Q&A 1:07:22 - end

The Developer Evangelism team at GitLab created this hands-on learning workshop for the Switzerland meetup. The generic template for future workshops can be found in our handbook: https://about.gitlab.com/handbook/marketing/community-relations/developer-evangelism/projects/#workshops

A

A

Great, so here we are, thank you so much everybody for joining. uh My name is adrian kosmacheski. We are on thursday january 21st, uh 22 20 21. So today is the 21st day of the 21st year of the 21st century.

A

So in a few minutes it will be the 21st minute of the 21st hour of the same day. So it will be a very I don't know if you are superstitious, you might want to play blackjack tonight anyway, um thanks a lot for joining. uh We are here on the first gitlab switzerland, meetup organized together between gitlab and vision, for those of you who might not know vision. Basically, this is how you pronounce the name by the way, the name of the company.

A

We are experts in everything that has to do with devops, docker and kubernetes. That's basically what we do. We take care of applications running on the cloud, 24, 7 support and, of course, we've got a whole lot of partnerships. One of those a very important one for us is gitlab, of course, and we also run our own uh container platform called a puyo based on openshift and we develop projects in as well to help us manage all of these systems.

A

But in order for enough about vision, uh the agenda of the day, it's very simple and very complex. At the same time, we are thrilled to have uh michael friedrich. His developer evangelist at git. Gitlab he's got a whole lot of experience about github. Excuse me, good luck, very bad, very bad lapses. About gitlab and um I'm. We are thrilled to have him on board and he will give a whole introduction about gitlab and all of its capacities, and, uh having said that, I leave the stage to michael, so I stopped sharing and michael.

A

The stage is yours. So just one thing, while you're sharing your screen, everybody please feel free to use the q a box on zoom uh to post your questions uh as we move forward. Michael uh will basically answer them either during or after, but please feel free to participate. Thank you so much so michael. The stage is yours.

A

B

um Before we start I've, I've shared the slides with you, so you can follow along um everything. We do today will be live and you can try it out or try it out later on.

B

I'm really happy to be here and to have the opportunity to like share some some wisdom, some best practices with you as a starter for with git lab, and also like some, maybe some advanced topics around ci, cd security and and so on. I'm. I know that, probably after after a while 30 minutes or 40 minutes, you might get one might get tired of zoom or something around this.

B

So if you, if you have a question or anything around, it, uh totally shoot it in the in the chat, um and there are also like um there are exercises which, um which are at the end of the slide deck, which should encourage.

B

You should just take half an hour of your time later on and practice a little and get to know get led by by just using it and exploring what's possible, and um this is the right window a little bit about myself before we start um yeah, I'm a developer evangelist since march 2020, before that I was involved for around about 11 years in in open source monitoring development and backend engineering.

B

um I do love community support social media twitter. Everything else, I'm originally from austria, um grew up in lint, moved to vienna, moved to nuremberg in germany, where I'm still staying, but right now, I'm all remote with git lab.

B

You can find me as dns michi on on social media on twitter, everything else, and I, when I don't hack or when I don't do gitlab. I do love building lego models, which certainly often is a problem, because my flat is not that big, but yet it's still fun to build lego um and moving moving along with gitlab. um First off. I want to give a short, a short introduction. uh What we're gonna do um then, the I think arnold.

B

First, steps with gitlab, then a little bit about cicd getting started, jumping into a gopro check to practice a little then some outlooks or some trying with security scannings with gitlab a little bit about deliverance deployment and an outlook. And last but not least, I've also prepared some exercises for asynchronous practice um and we can totally try it out, live or would just like keep keep discussing uh discussing. What's all the else is cooking, because probably I could talk about gitlab 24 hours, but it's we just have two hours, but it's two hours.

B

It's still long, just for you. If you want to follow along with the exercises um from the slides, you need a gitlab.com account um and will the all the exercises will be done within the web ide and the editor. So you don't need any shell access or terminal around this um first things. First, this repository is public and we will be working on it throughout the workshops and the exercises, and you need to fork it into your own account and then you're good to go.

B

um You can also use the presentation to slice the exercises for your own workshop, so for your own ideas to spread. The word, explain things to colleagues to your friends um and even getting inspired from like best practices or things you you learn from today.

B

Okay, so this is one of the uh what is gitlab overview. Slides gitlab as a company was founded in in 2014, um we're happy to have many open source contributors. So there is a community edition and an enterprise edition, and basically it's it evolved over time from just managing source code, which which is kit um into the whole collaboration suite um and the devops lifecycle, in a way of like from the first code, to the production deployment and everything around it with reporting monitoring security and whatnot.

B

So it's um it's growing fast in the market, um some more technical insights into gitlab- and this is just a summary which I brought up over the years myself when I was creating gitlab trainings in the past, um you combine communicate with your git client over https or ssh.

B

You have the gitlab interface, which is either the rest. Api is an https endpoint, which you can use for literally literally anything which means, for example, creating users, doing monitoring ultimate tasks which could probably be done manually otherwise, and you have, of course, the user interface for everything else and deeper collaboration and development from the backend. You probably won't see anything when you're using the self-hosted omnibus packages installer, but just getting getting a better picture of what technology is in the background.

B

um So basically it runs. It uses ruby on rails, so also for, if you, if you would consider contributing to gitlab, um knowing ruby on rails, definitely helps um it's using pumas as the default web server with nginx hostgrass curl is the backend um red is for caching, server-side rendering is done with nodejs and all the daemons which take care of specific high-performance tasks are written in golang, for example, the gitlab cicd runner, which we will learn about today. um Gita live for git cluster management and also the release cli and so on.

B

um All these components are fully packaged into into omnibus, which is kind of a shift recycle in the background, and it works during upgrades um very very nice. So you don't need to care about. Like does the database upgrades work? No, um but instead you need to be prepared um for the first steps with gitlab. um I would totally recommend to just like, for example, start with the dashboard get an overview.

B

um What's in there explore the menu header with the actions um figure out. What's what's in inside um you might. It looks familiar to github and and um and other um web uis, I would say, um but it's it's the main entry point and knowing where to click and know and knowing which, for example, keyboard shortcuts are being useful, um can totally help. The other thing which is which is important, is in within your profile. um You have the possibility, of course, to set a nice picture.

B

um You can also add your your your biography and the most important thing is uh within your settings. um You can manage uh all the access stuff, which means, for example, adding your public key for ssh or creating a personal access token for http access, um which you will need for anything for any integration, for example with visual studio code and the gitlab workflow extension or anything else.

B

um Another thing is everything which means you want to organize something um aside from the projects hosting uh the git repositories, you can create multiple levels of groups which makes the organization easier um like nested sub groups are possible and the good thing is: if you need to find granola permissions, you can actually do that from the group level or from the subgroup level and saying hey.

B

um For example, the main uh community relations team has, I think we have developer access or maintain access, but then specifically, I can grant, for example, ownership to the evangelist program in the screenshot. um The other nice thing is, you cannot mention a grouping issues, much requests and comments, um which means you don't need to remember whom you want to tag just like add and typing the group name autocomplete and then you're done.

B

Another thing when getting started and getting familiar with like planning your gitlab instance and and overviews, are the issue lists and also the issue boards, especially because you can have quite a lot of issues over time and you might see it on the gitlab project itself, where we have, I think, 35 000 issues um and we don't really look at the issue list on its own, but we have we use, searches or actually filter, filter searches, urls issue boards and other ways to look and plan for the future and the future releases.

B

So this is a good way to for once practice the filters and then later on, store them in issue boards, kanban boards, road maps burn down, charts and so on.

B

um Yeah and another thing I want to highlight- I'm not doing the lab right now, it's just more of an introduction. You can of course do it live as as with as I talk along um using issue templates- and this is this- is one of the things where you don't want to play like ping pong, when I'm opening a bug report or an issue, a feature request of future proposal. I want to get an uh get a list of instructions uh what's what's needed to reproduce.

B

The problem um to, for example, get some get some ideas.

B

I'm saying hey this feature looks in a competitor product uh like this, and we won't totally want to do it like this and that some some visuals some some designs and whatnot, and this is easily doable with so called description templates, which can be used for issues, much requests and also epics within gitlab, and it makes it easier to collect um for once, feedback from the own development team or, if you're, looking into opening up the repository of open source contributions, it also helps um other than that um next to issues on its own.

B

You want to organize it in probably in a roadmap, and one of the ways you can do so is by using milestones, so you can um create a new milestone, use a semantic version and versioning system mapped to the later on release and assign issues to those milestones.

B

This is one of the ways you can organize this and another thing of organization.

B

With this issues are so-called labels um like you, can use them for filters um in issue boards or in issue lists, but also they have a visual helper thing around, which means you can can immediately spot the bug, because the label is red as a background, for example, so they have a signal color thing inside um the other thing is, you can also use like prefix or even scoped labels, um meaning to say you can define workflows, for example. So we have uh labels around uh pro um we're getting a request.

B

Then, at the record request is new later on it's open, then it's like in progress and and then uh it's closed down, not done not accepted or whatever is needed. um Gitlab provides a default set of labels, which is which is nice. um One thing I would recommend this or which is also possible. Well, I would recommend you can also define a group level uh setup, a group you can define uh group labels, so the labels are then inherited to the project and you don't need to manage them on on each project.

B

um The other thing- and this is one something which we will practice um in the next minutes. um We have the uh the plain or the static side editor and also the web ide. The web ide looks like visual studio, gold or some other editors, um and it's um it's super easy or super convenient to just like fix the typo in the documentation or in the code edit.

B

It create a new merge request or work in the in the in the web interface in the browser avoid any context switches um and you don't need to clone the repository. You can just work within gitlab.

B

This is one of the things the other thing for especially for starting with gitlab is checking out the project activities. Sometimes it's a little bit hidden because there are many features, um but the uh you can view the global activity like on gitlab.com um or you can also go into the project specific overview and say hey. I want to check out this activity.

B

I've I've lost something- um or I didn't remember, what's going on um so there's there's a way to look into that um and you can also see what's what's going on with uh gitlab.

B

As a project for.

B

Example, can you mute yourselves, please.

B

um Yeah, last but not least, uh we have the to-do lists um every time you, for example, attacked um so you had mentioned, um or you have you get an assignment of a much request or a ci cd build failed or something else around. This is like a to-do for you and you can then say hey it's done.

B

You can filter by it, and this is basically one of my workflows um on at gitlab when we use gitlab to create gitlab um just to like keep everything in a single place um and I'm barely reading my emails, for example.

B

Okay, um now from from the basics, with gitlab as a management system and project management and source code, um the core part of one of the biggest parts is cicd, so continuous integration, continuous deployment, continuous delivery um and just to like get a better picture, get an overview.

B

um There are some like key concepts uh to understand like a job in a ci cd pipeline means you want to execute a certain set of commands which produce some output, some exit codes um and like the overall state, which should be presented later on um the pipeline groups, uh groups those jobs into specific stages, uh where you, where one job triggers another job and so on, like similar like you, would be creating, for example, um a car and the manufacturing goes on from like the basic skeleton to the doors and so on, um and you also want to ensure that, for example, not there are not like two left doors being built into the entire car.

B

So at a certain point of failure, the jobs should uh cancel the entire pipeline. um Another thing to know is that there are environments um for specific uh deployments. For example, you could use a staging environment, a deaf environment, production, environment um and in order to to instrument or parametrize the jobs, you would, for example, use environment variables variables, which can be helpful to say, hey. I want to trigger the the pipeline for my ios builds, or for my android builds or like specify a specific software version which should be checked out.

B

um You can also hide uh credentials inside environment variables, which are then being edited in the project settings or in the group set uh settings hidden away from public access. Of course, um yeah.

B

The the aforementioned outputs are called so-called job artifacts, um anything which can be useful, be either job log output or a metrics report security reports, binaural packages, starboards anything you can imagine actually, last but not least, the ci runner or the gitlab crc runner crm. It's a historical thing um renaming that to ci cd. um This is basically the the engine or the power hose which actually executes the jobs and communicates with the gitlab server.

B

So in in theory, you could, for example, install the gitlab runner written in go on on your macbook on your windows, desktop or just like.

B

On on linux, the other thing to mention is: you can also use auto scaling with the runner, which means you don't need to run all the resources in your cloud environment, all the time which would totally explode your your total costs in a month, um but to say hey when there is a certain bottleneck or there are too many jobs in the queue currently, um I can just say like in hetzner cloud or an aws or whatever I need it um spin up new virtual machines, then the jobs are getting executed and when they're idle again shut it down to specific limited number, um the other way to scale it is to use kubernetes and instrument and orchestrate the containers there either use it with hosted, kubernetes or managed kubernetes, which is available, for example, in aws and gcp in azure, wherever you can imagine or use it in your own, managed, kubernetes, environment or project or whatever you want to use.

B

um The other thing to mention about gitlab ci cd runners is, you can assign them dedicated to a project. A specific group use use then tags or have them globally in the pool which could be, for example, shared runners, as you can see, on gitlab.com.

B

And this is this: can uh can get uh handy when you, for example, have on the left-hand side, see plus plus projects which needs lots of resources for compiling and building um and, of course, more resources. Then, um and on the other side, the front end is written in, let's say php um and it doesn't need so many resources to build the code and to compile it.

B

You don't compile it um to like, create a release package, for example, it doesn't need so many resources and you can use tags um to assign the runners where needed um other than that um for the for the naming the ci pipeline, you commit something, you run, you build something you run unit tests, integration tests, which is basically what we call continuous integration, and then you have a sort of delivery, so you, for example, prepare release you automatically deploy deployed in production and um from the deployment standpoint.

B

You probably want to do certain reviews, aided by revenue apps like deploying the current git branch into a kubernetes cluster and reviewing it if everything works. Fine um in a staging environment and production um is like a screenshot of how a pipeline could look like, um and there are way more things than I can explain in in this short amount of time what you can do with uh pipelines. um The simplest iteration of the simplest way to use pipelines. Is you define like build test stage in production?

B

You have a synchronous pipeline which is blocking, but there are other ways to to make use of advanced pipelines with, as in chronologist dependencies like the directed alcyclic graph, you might encounter throughout the docs.

B

There's the possibility to have multi-project pipelines. You can have parent child pipelines, pipelines would chan with generate which generate ci configuration code for other pipelines which are then executed. This is something I call pipeline inception or, for example, merged range, which kind of assume or how to explain that you have. You have lots of changes running into your main or master branch um and oftentimes something breaks or fails and merge. Trains are basically like a queue where it uh simulates. What? If all the all the much requests in the queue have been merged and mine?

B

On top, would it fail or not, and I get the result of that- there's a blog post out there written by vitica.

B

I need to look up the url later on, which totally explains that nicely, but anyhow um uh for getting a little little deeper into like actually practicing um gitlab and gilleb's icd, normally or typically, when you start working on a new in an on an issue, you create a git branch.

B

You commit your changes, you create a merge request which can be a draft of in progress so that everyone knows that you're working on this, then the ci pipeline executes tests or you get immediate feedback during your development process, not not just when everything that gets merged later on um and you can, for example, react on code quality problems.

B

There could be certain security tests um telling you that you probably leaked an aws credential in your unit test right now um and at a later stage, when you say okay, I'm confident enough. I want to ask for a review assigning it to a to a colleague.

B

The revenue process starts peer review discussion, everything around it. um Someone approves the changes and it gets merged back to the main master branch. The ci pipeline runs and deploys it, and even though you want to, for example, monitor the application, which also can be done in the merge request. For example, and you get an idea of oh, this git commit actually increase the memory usage by a thousand percent. um We, for example, cannot merge that um this is also not a possibility.

B

You can implement into your ci cd process or pipelines, um but I'm talking too much. We need to do something. um The first step with gitlab ci cd is- um and this is what I would recommend um just like keeping things simple and the most simplest example is to just I want to spawn a job with a new container image um we're using docker, and in that regard I want to use alpine, which is, I think, 20 megabytes of memory footprint.

B

um The colon latest means just using the latest tag which is available, um and then there is like the free form for shop names. So there are certain keywords: on the on the first level in yammer in the in the configuration but, for example, the job name test underscore symbol is three four and the chop has a requirement of executing a script so oops.

B

This is a script and it should execute something and the dash is basically a yaml array or yummy list, um and this is something we want to do. I want to try out now so um in order to do that, I'm switching over to the repository, and I want to use the web ide and just waiting here. I want to add a new file.

B

This is basically in over here um and I want to use the dot gitlab ci dot yammer, because I'm lazy- and here I want to use image and it does have auto completion.

B

I want to use alpine latest and the job was called. I think test simple and nice. Nice thing is that the uh the indent works um and we say exit one so- and this is basically it over here I have the possibility to commit it and say add ci config.

B

I want to create a new branch which I call ci config and I want to create a merge request. So it's not immediately pushed to the main or master branch, but instead pushed into a merge request.

B

And this basically looks good, I'm more interested in showing you how the pipeline works. So in this example, we can see that the pipeline is being run and when I, for example, click on this, I can navigate into the pipeline overview and then see test simple.

B

This is the job which is being executed and it did something. So I pulled the the alpine docker image prepared, the repository and called exit one x.

B

One is an error code, so everything above zero actually is an error code um whenever you use that um or when you execute something, and it would produce an error code or an error, and you just want to capture the the output, for example, for creating some unit test reports or some performance tests, you need to ensure that the right exit code is being returned because otherwise the job would fail.

B

In that regard, when I want to navigate back to my merch request, I have it linked over here and to show you that it works. I can go into the web id um and in that example,.

C

B

Want to use the diff view, but just edit it and say hey, let's just pretend that we're going to fix this and fix see. I config committing this and then I can see, for example, over here at the bottom, um that the pipeline is being run and the nice thing is. It opens up on the right hand, side um and while I would be, for example, continuing working on something in the web id I can totally inspect.

B

um What's going on over here um or just click here and say, hey, I want to see it in the bigger picture and in that regard it was exit zero. Everything is green. The first steps worked.

B

Moving back, this is the second exercise which we did already now. um Another way of dealing with uh with the ci configuration or the icd configuration is um to use variables um to instrument or to to define a different behavior within the jobs and the executed scripts, and this is one way of doing that. The global variables keyword can be used within the ci configuration you can also specify them on the job level or, if it's, for example, credentials or something else, you want to mask in the outputs which should be a sensitive information.

B

For example, you can specify them in the project settings and only the owners have access to that information, but, for example, the developer role does not have access to that typically example would be, for example, mysql root password for testing something else in that area.

B

Yeah, this is one of the home exercises for you, because I want to jump into actual code right now and um I'm gonna see how we can jump start um the gopro project. um The thing is um the repository which we're using for for our exercises contains some go code. There is a main.go as an entry point. um We have some logging inside um and anything else.

B

The main idea is, for example, if, if you're starting with gitlab grcd, maybe you have imported it from svn from a migration or you got the code from someone else and you're not really sure if it works or if the unit has performed or something else. So the first entry point- um and I do the same when I find the project and it doesn't have any ci cd around it. I I want to add gitlab cicd to it and see if, for example, the code is building dependencies are fulfilled and anything around it.

B

So um the first step in our go project will be now to try to build the go code, and this is like the verbose instructions. What we're going to do now? um We will be adding um or creating the gitlab ci configuration again, um and this time we will be using a template um not to avoid to type too much and editing some uh things like the wrapper name and also replacing the my binary uh thing with tanooki and then we're gonna commit it and see how things go.

B

So I'm navigating back into the main repository and we didn't merge the merge request now, but just again, fresh repository, we're adding a new file, um calling it gitlab gi and in this case I'm navigating over here, choosing a template and saying: hey, I'm looking for go and I just want to use golang and I will dive into the details in a bit. I just want to want to make it run. One thing we should change is the ripple name, um which is I need to like use this one.

B

um This is the repository name and then your fork. It should be a little different. I think it doesn't really matter, but I'm just doing it for the for sanity. um This will create a sim link, and this week in a separate workshop, I ran into the problem, avoid any trailing slash. Slashes um this looks correct, but I kind of removed this one um without the leading https, of course, and my binary is over here and replacing this with tanuki and this one as well.

B

Okay, and once I've done this, I should be good to go. um I want to add ci config for go project.

B

And this is ci dash, go and creating a new merge request.

B

Again, uh you know the special guest is fine, submit it and I want to inspect the pipeline, um and while the pipeline is running, I can also open it up in the web. Ide again.

B

I don't want to defu this one, so basically what it, what it is, what what is going on it prepares to go path. um To be honest, I cannot really explain it. The most important thing is, um there are stages defined, so we will see specific jobs um later on um and we have the test build and deploy stage, so they are executed in this order and chops can be assigned to a stage by using the stage keyword in the in the job. So format is a job and compile and it executes a script.

B

So this script, for example, runs. Go format, um does some other syntax checks and then actually calls the test instruction and the compile stage, for example, builds um aesthetic binary, which is output as tanuki, um and since we might want to inspect that um the after effects keyword defines everything which is defined as a drop.

B

Artifact and job artifacts are uploaded by the gitlab grinder to the gitlab server, and we can access that from the web interface, for example, or we could totally use the rest api and some automation and download the artifacts from chops, for example.

B

um Okay. This is in my much request, which is linked over here, and it failed. Oh okay, let's probably inspect that and see where it failed and like we can see um the format shop failed and the compile job wasn't even triggered by that. So it failed fast, which is a good good sign and it the formatting was fine, but the dependencies of the includes were didn't work and in this example uh it's in main.go and line 12..

B

So we totally should check it out um and there is something broken. If I never get back into my merch request now and open up the web ide again, um I want to go into the main.go.

B

I have set line, number 12, which is os, dot, get ends, um and it's obviously it's intended to break um because the inputs are missing and in order to make this exercise as easy as possible, um we can just remove the uh commented out this code import. The fmt uh package, which is used for the print line and os, is retrieving an environment variable variable, which you can, for example, pass from the cicd variables as an exercise, and this should yeah. This should fix the source code.

B

Let's commit it into the same branch and saying fix missing, go imports um and committed wait for the pipeline to go on, um and hopefully it builds nicely all along um if you run, for example, into scaling issues and the job takes ages, for example, to pull in the docker image. Maybe you've hit the rate limit, for example from docker hub.

B

um There are ways to to optimize that in in a specific sense of either using a local registry uh using the dependency proxy for caching assigning more resources, auto scaling to the gitlab runners, you can also, for example, use a gitlab.com account and register your own runners. So, for example, yesterday in in our everyone can contribute coffee chat.

B

We tried uh hits my cloud with auto scaling, which also worked nicely indeed, in this regard, the job succeeded, so everything was fixed and I'm expecting or can expect that the pipeline is executing the compile job and this again passes over any.

B

Any things yet now it's a different stage: yeah it should it should pass the artifacts, um and this is typically the time where um someone gets a coffee, which I totally have.

B

um Okay, the job succeeded and it generated the the tanuki binary and uploaded it to the coordinator, which is the gitlab server, and over here we can see. um There are trouble effects unknown coffee here, fetch the beer yeah. I totally agree on that. um It's it's after 4 p.m. uh The tanuki binary is uh over here and if you, for example, on a linux host, um you could also execute that. I cannot do that on mac os, but I could add a cross build or something else.

B

um The thing is, I don't want to do that. I want the gitlab ci cd to execute it and see if it check for so.

B

If there are any problems at runtime, and one of the things I can do is to navigate back to our exercises- and I want to add a new job which should be run, which is called run, should be executed in the deploy stage, simulating a deploy or production run, and it should just execute the tanuki binary which was produced by the previous job, and but in order to do that, we can navigate back into the web ide and ensure that we're inside the merch request.

B

So in order to do it from a fresh start, I'm gonna go into uh the much request which is on this one and it's great to see that everyone is practicing um going over here into the web id again, and I wanted to defeat this one. Adding a new job.

B

um Let's call it run stage is deploy and script is dash, dot, slash, taluki,.

B

Okay um and then again, we can see we have. We have one uh another chop which is being run um and this triggers the pipeline again. um So it's like you have um you don't have any like mixed up together: testing environments, it's like just yeah, you have the first job, it runs again and again and again um and you you get immediate feedback. If something is broken um in that example, um it it runs and runs. um Let me just see what else um is coming yeah while grants.

B

There are other possibilities, you probably want to use within ci cd, and one of the examples could be running unit tests unit tests. I think we have too many programming languages. It differs all along, so there are different frameworks marks things to do it go. Has it built in nicely done, and it's also possible, for example, to generate coverage reports, so the test coverage and then use the trade unit integration in gitlab, which passes the xml report and provides a nice overview within the pipeline view where you can see.

B

Oh, the tests have been run which which ones have been failed and so on. um The unit tests are also important, um for example, when you have a merge request, workflow and the tests are failing. This is a sign for the reviewers to just leave it there and say: hey. The developer knows got a notification, email and the notification to do in gitlab itself. They need to fix the test and once once it's ready, um someone else will do the review. For example, um this is a way to ensure that resources are being added efficiently.

B

um Let me just see if everything worked out now. This is the wrong top. This is the right type. The job is executed nice um and, if it executed nicely, we should see the tanuki our input tenuki, which is printed over here. So everything is done. um If you see it too, I would love to see it on on social media, for you to share um tech, gitlab or tech dns me um yeah, and so we can enjoy what's possible. You can also send, for example, improvements to this repository or something else.

B

And this was for a ci cd, um the other thing, and this was basically the exercise for child, which I did 30 minutes before the workshop. um I wanted to show you something which is new and which, which we totally would love you to try out um it's a new pipeline editor from the accessible from gitlab from the web web ui.

B

It's it's live from gitlab.com and it should be out in the first iteration tomorrow when 13.8 is released on the 22nd um and in order to not just look at the screenshot we can uh use. Let's see uh navigating back into the repository, you probably have seen it already.

B

um There is a file called uh dot gitlab.ci.solution.yammer, which contains everything which we are doing today and and a little bit more than that um in as a solution for you to like verify what, if everything worked out um and we're gonna use this solution to create an initial working, uh gitlab ci oops in the kit, let's get lab ci configuration. This is currently needed for the for the editor, so um shouldn't be too much to worry about.

B

The easiest thing is- and this is like an advanced uh pattern you can use with uh the ci configuration you can use includes and, for example, include. A local file which is gitlab.hdis, cannot type today solution, yaml, for example, you could also import existing templates for code scanning ci security.

B

You can define remote files from a different, https, endpoint and so on, and in this regard I just want to include this and push it actually to the main branch to have it available in there, and um this is just a way to show it to you.

B

We have it now. I just want to get back in the main review. um The pipelines are being run, but instead I want to navigate over here um open up the editor, and um this is the first iteration. The most interesting part is actually a visualization of which jobs would be executed and probably um you're wondering why there are so many test jobs right now. um The thing is there: are security templates included and they generate jobs for you, so you just include a template and you get everything which is needed for security.

B

um This is under heavy development. I would say, like everything is under heavy development. um This is the first iteration um and we're totally planning for more more. um I need to find the um the epic where we are asking for feedback, um maybe just like I'll just do a. I should hop onto twitter onto my profile because I have the tweet pinned and there is the epic.

B

um So I can share it with you uh for feedback, and um I would totally invite you to try it out. um Let us know what you think: what you're missing, um yeah and also the nice thing is um it has life lending, so you don't need to commit and wait for the.

D

B

The yama era, which could be totally like you, cannot drink as much coffee as you want to um okay. This is this example, and um just for the sake of completeness, I will reset that later on, um but one thing I also wanted to show you, and we already did it now- um is to like what can I do about ci security or security in general, for example, to detect, detect any secrets, um for example like an aws, credential or specific password, or something else which could be hidden in your in your git history?

B

This is, for example, a way to to you make use of gitlab and the other ways secrets is one thing. The other way is, for example, static, application, security testing, so detecting whether you use chmod 777 for some reason on the file, or maybe there you are not closing any file descriptors, which could lead into potential um stack overflow, whatever attacks um and yeah.

B

Basically things you probably do not expect, even with like dynamic application, security, testing, cables, engineering and and the like, um and I think, let me just see in the uh repository right now. Do we have a security? Oh it's over here in our security dashboard.

B

Do we have a vulnerability report right now and that's the list now? Okay, um I could open up.

B

Oh what did happen. The pipeline failed for some reason: ah yeah, of course. uh Of course it failed. um We didn't merge the much request now. um Okay, then, let's probably I shouldn't have added the exercise last minute. Let's fix that live um go into our go project again and edit. The web ide edit, the ci configuration and um yeah just do this again.

B

What I did before um include you could also use a shortcut and just include it like this, um even though the the web ide doesn't like it uh include this over here, um and it will security jumps, and this should be should being around then um go over here and.

B

Okey-Dokey, we want to view the pipeline again and waiting while this is being run. um The screenshots show the integration into the much request, and this is currently available in gitlab ultimate um in the enterprise version. If you, for example, want to automate it or use it in the lower tier, you could also use the edit effects for the json reports, which are generated and, for example, trigger the api. Add a comment into the watch. Request build your own report system.

B

This is totally doable and while this is being run, we can just hop over what else is possibly interesting, and I think I'm a little over time, but so much things to tell um all the defaults is something you probably want to try out um it's an opinionated um set of ci templates abstracted away. The idea is that you don't really need to care about dependencies for python for go for anything else.

B

It tries or it attempts to build everything for you deploys it pushes it into the container registry and the first iteration worked with kubernetes workloads and, most recently, aws has been added to autodeploy.

B

I think in one of the recent versions- um and you don't have to like look into a specific ci configuration that much um the other thing is like everything you can use for a security testing um like secrets, testing and static application. Security testing would be something I would totally recommend to enable as default um dynamic application. Security testing is where you, for example, penetrate and rest api with like malformed input and try to trigger memory, exceptions or something else which breaks the application or even leaks.

B

Something um dependency scanning um is also one of the big parts where you want to explore, um or you cannot really control when you're using nodejs, for example, and the reverse dependency included a cve and exploits your entire production system. You totally want to prevent that and say hey whenever a dependency is being bumped or it's detected. Having having a vulnerability, um you want to get feedback um on that and fixed.

B

It um container scanning falls under under the same like direction: um oftentimes you're, using a base image which gets vulnerabilities and cvs assigned, and you want to detect that, for example, using dbn buster in the 0 release has certain kernel problems. I think um it's totally totally normal, but you need to take care about this and be aware of possible problems in the underlying docket image. For example, um fast testing is totally new, um which means, like chaos, input chaos, engineering, no human tests um and everything around it.

B

I would totally recommend to explore it um and before I jump into what's next, is it being run? um It's not not entirely complete yet, um but let's inspect it, security scanning is loading. Maybe we have a chance to look into it.

B

Can I view the full report?

B

ah The full report is more interesting, so in this case it detects an awa aws key, for example,.

B

This one is, I want to have more info on that and I can see, for example, that's in the dot n file, so I obviously added a dot end file where it's inside. I can, for example, click on that and see that this line is affected. um One thing I can also do is, for example, create a confidential issue from that, so that someone needs to take care about it.

B

This is also possible, and what else do we have we have? I think you want to go back. There is something around yeah.

B

I added a method which does something which you normally don't do, and this is also like to to detect certain things um which which could potentially uh like leak credentials or permissions, or something as around it. Okay, um this is about security, um and so so to speak. What's next, what's what's interesting even more, if you want to go advanced or want to say hey, this is important for my development workflows for my devops journey.

B

um You can use the environment with staging and production, look into review apps what could be useful for your environment or your for your application for your development? What you need, the other thing is git ups, which means like infrastructures code, managing your either kubernetes looking into terraform as an automation, layer or even ansible.

B

Monitoring is always a big part and observability. um You totally want to like when you a pipeline is always similar. Like you write code and you reflect the code, you improve it. You learn things um and the same goes with pipeline efficiency, which is linked over here. This is a documentation which we wrote last year or added last year, around optimizing, the pipeline runtimes failing fast, often oftentimes, optimizing, some caching, the docker image size and so on.

B

Incident management is also a big part which we are like, um adding the possibility to receive alerts from external land points providers and managing the incidents directly in gitlab and last but not least, next to all the security stuff um for cont for code scanning of for the code itself.

B

The ops team or the ops part should also be mentioned, with managing secrets, for example, with hashicorp void and not hard coding, any credentials or secrets anywhere else.

B

Other than that, this is basically what I planned for today. The rest which I will be like skipping over now for a bit is something which I would recommend for you to practice by yourself and let me just go over there and you probably have it open already. There are some tips around much requests how this can be optimized or how you want to how to use that. There is also a way of like using uh push options um in your cli totally interesting.

B

I did I wish I would have known that five years ago, um other than that there is uh practice how you can automatically uh close an issue when a merge request gets merged. Super nice workflow um also like reviews, work with suggesting things um so when you're, adding editing something you can inline make inline suggestions and the author can apply it or someone else which makes the workflow more convenient and adding this to um and apply workflow and the proof workflow.

B

So when everything is working out- um and there are some tips inside with some checklists in the much request template if you're looking for real-world examples, please check out how we do it at gitlab. We have certain much request and issue templates around which are long but they're, useful yeah when everything works out practice.

B

How you want to do releases, how the workflow is, um how you feel confident around this around this um and um it's jumping from the releases back to a test, a little bit um test coverage reports for you to practice um with some more detailed instructions, how this can work and um what else do we have? Oh, we have actually examples how to add a new function and then add tests um and see how the test coverage evolves over time. And last but not least, we have the batches.

B

So you don't need to edit the readmemd.

B

um You can just go to your project settings and add the svg integration um yeah and last but not least, there are some certain tips from which I get asked all the time, and so I decided to add them over here as well, like git branch, best practices from from my experience over the past years, with release branches, support branches and so on um and other than that, if you want to, for example, jump into um personal access tokens and integrations, oh, and I think I made it okay, I copied that those slides from a kde workshop I provided around gitlab last year.

B

The gitlab, the the personal access token, um can be defined for specific scopes, for example, so, like only having access um to the rest, api read, write or read, um because the reason why we want to use that is, we want to interact with the rest api and there are certain api clients for specific languages and libraries around I'm a fan of python of the python gitlab library, but there are certainly others around super convenient not through having to write the json requests and http requests on your own all the time.

B

It's just wrapped around other than that girl is a cli tool developed by our amazing community, called g where you can manage much requests intact with the code and so on. um You can install it and more tests and reports, something.

E

B

And last but not least, everything which is about efficiency and ci, analytics and monitoring some more insights um and also the ci pipelines exporter, which could look like this in your grafana, for example, if you're using uh promises and and the exporter.

B

um This is also linked in the efficiency docs.

B

The other thing is like failing fast, so when your ci cd is blocking, for example, the code qualities can run seven minutes in my tests, um but the format job doesn't work, so it makes sense to to do it in a different way and optimize. The the ci cd cost um yeah, and I think I have listed too many tips and tricks here. um The metrics builds for parallel job execution has been added a couple of releases ago, making your life easier as well.

B

um It can be totally useful if you, for example, want to generate rpm on dbm packages for different architectures, um just by providing an arraylist of that or a matrix.

B

Directed is a click graph for speeding up the critical path of your execution pipeline. This is a thing to look into and also everything which can be optimized with uh container image creation, um creating docker images, which already include the build chain um which got better over time, so you can actually use. I think I think there are certain dbm images, for example, which incorporate the build chain with gcc, so compiling c and c, plus plus code, for example, and so on.

B

um Container registry is also different thing to keep in mind document and share everything.

B

And yeah produce useful errors um for anyone looking into the cicd job, output and document, it document everything. That's that's good when you know where you want to look at yeah with that, I think oh one hour, um thanks for your attention, I would open it up for um for questions. um If you need anything, you can like tag me on right. You can tag me on twitter. Write me at the m, looking forward what you, what you learned or what you have which questions you have other than that um yeah thanks for attention.

A

Thanks a lot michael, it's been a fantastic introduction to gitlab and I I I actually learned many many tricks that I had never. I had never seen never knew about. So that's really really cool. So we've got a few questions, so there was one that arnold answered uh during the talk, which is how much how much resources would I need to run my private gitlab instance in my cluster and arnold referred this question directly to the documentation.

A

So it's four gigabytes of ram recommended as a minimum. So that's the answer.

B

Yeah yeah, it's. uh I would also point to the docs. To be honest, exactly gitlab is an all-in-one experience requires certain resources.

B

um Give it give it four gigabytes to make it happy um if you're trying to scale it up um like in in a larger environment, and then I want to use, for example, a distributed uh environment with different instances all over the globe or maybe like yeah. You have 10 000 developers, maybe uh or even more than that, give it resources we have.

B

We have ongoing projects to optimize the performance of gitlab itself and also, for example, puma reduced the memory usage and things around this. So it's it's recommended to give it as much as possible, of course, but um for for starting, I would totally recommend for gigabits, perfect.

A

That's good, so there are two open questions very interesting one. So the first one is: is there a way to include a mandatory gitlab, ci yml file in the ce edition, for example? If I want to have a certain stage mandatory for all projects in a group, is that something that it's doable.

B

You mean to always have a ci configuration file available for all projects.

A

For example, if you have your group, you want to enforce a certain set of practices or testing, or I can imagine a few scenarios where you would like, as a manager to make sure that all of the teams contributing to in a particular group have a default. Let's say gitlab ci yml file that it's automatically included in every uh in every new project in that group. That's how I understand this and I actually could imagine a certain usefulness. Yes,.

B

Yeah, definitely um I'm just looking or just googling. At the background. I know that you can, for example, disable um or you can the other way around. You can define a remote gitlab ci configuration file in your project settings which then will be used.

B

um The thing is, I need to find the documentation um and figure out in which, if it's a white label, uh a second gitlab, so um I'm a developer, I'm googling everything I just need to be fast, absolutely project settings and, to be honest, gitlab has so many features and with me, starting in march, and before that I've been learning and providing trainings, um but there were still uh like questions like like this one or say: yeah. I've heard it somewhere, but I I need to look it up so.

A

I have the person magnus lubeck the person who is asking the question raised his hand, so I'm going to allow him to talk. So maybe he can explain directly the purpose of the question uh magnus. You should be able to talk right now.

F

No hello! No! No! I think you can hear me with me. Yes,.

A

F

Hear you yeah, so I read about this uh feature uh in in the licensed version. So I think there is a quite easy ways to make mandatory that certain pipeline steps are uh run before your own cic get. The fci file is a run used, for example, for compliance reasons. I think it's very common in bigger organizations like banks, insurance companies, that a developer cannot skip certain steps that used to have to be wrong, because the github ci file is, of course, editable by more or less anyone.

F

So that is sort of the question that uh for some open source projects, where you can't afford the paid version you might want to have- uh maybe not for compliance reasons where I can restore understand that uh you would like to pay for it. But for your own hobby projects that you have your your favorite pipeline step that you would just like to include in every project that you have.

B

I think so one thing: if, if you're, if you have an open source project, you can totally apply for the open source pro program and get a get a free license um of the enterprise edition. This is something you can do. The other thing is for compliance reasons, and as you mentioned it, I I couldn't find it right now, but I I remember discussion around this. um I think it it's available in the project settings.

B

I think it's uh it's. The customs yeah configuration pass, probably um approach it if needed, and let me share my screen again.

B

It's easier: this is the correct one.

B

um Oh there's the share button, um I'm thinking about not this one, it should be using uh okay, it's moved away um the custom ci configuration path. This was added uh introduced for external locations.

B

They should be able to override this. um Let me go over.

F

B

F

Super important to find the solution. It was more if you knew it from the top of your top of your head and then because I know the feature exists in the in the paid version.

F

It's just that I don't know exactly if it was available in the free version.

B

uh Sorry for doom scrolling, um I just wanted to figure out um 12.6, because I remember it from I think this issue or something around it would just copy it in the zoom chat. um I think you can forcefully override this. The only thing which I cannot answer right now is in which version it's it's included.

B

Custom configuration path.

B

We can customize the past settings. Cicd provided custom safety changes.

B

B

Inside over here, um probably, if you haven't done so, try this one out.

B

But yeah, I seen a comment in the chat that probably it's in the paid version, and I I just cannot figure out how the feature is called right now. Oh.

F

I'm very happy with uh this. I also agree with uh simon in the in the chat that, of course, in uh in community uh projects, you don't really want to have someone used to whipping people's houses was more more for a personal preference. If I have my own favorite uh steps that I always want to include.

B

I'm thinking of, if you you know, maybe you could.

B

One thing you probably could automate it in a way of saying: hey, you have your your template group or your template project somewhere, which enforces that and is publicly accessible um and then you're using cisco use, which is a crown job actually, and you kind of check why the api, whether someone over tried to override this um okay yeah, it sounds a bit workaroundish, but you can sort of like do some compliance by yourself.

B

You could also like write a ground chip which checks the rest api every five seconds um to ensure that the specific setting is not overwritten. um It's also compliance, but not it's not how you sell it to someone, but you can like make it happen. Yeah.

F

In a private project, I can use the use and include in every uh github ci file. If I want to.

B

Yeah, you can, I think, yeah. If you're working with roles and protected branches, um you could also like say: hey um only trusted members can merge to the master to the main branch, the default branch um and within your merge requests.

B

Yeah. You need to define a check um that no one can like modify the gitlab ci configuration file. um So the review process includes that there is a certain check being run um or a task list in the merchandise, template and saying: hey you you're not allowed to modify this and no changing. That regard will be accepted from an from a community contribution. For instance, um we probably I don't know how we handle that, um but I don't think we would like watch something which changes our ci configuration.

B

So this is. This is also like training, your country contributors and defining a process in the in, like the merch request, reviews and saying: hey, um don't fix everything or touch everything in a merge request, only what you want to fix and if you want to adjust something in the in the ci configuration.

B

Please create a different merge request and this gets assigned a specific reviewer which, which then says: okay, I'm trusting this, and this can be added.

B

Oh cool just take something I could think of.

A

We've got another question from nicola jam. I don't know how to pronounce your name. I'm sorry if I mispronounced it, where are templates included from, can I browse them and can I freeze the version of the template in my repo.

B

That's an interesting question.

A

um I've just given mikola access, he can. He can basically talk if you, if you would like to expand your question nicola, please feel free to intervene.

G

Hello, everyone yeah, I think I've written everything I was going to ask so first part is like it looks a little bit like a.

B

G

We have this included template with some files that don't exist in my repository and it works and second part about freezing version. It may be required. Okay, I can. I can think about some cases that may be required like if we would like to have the control over those updates.

B

Good idea, this is something I don't know um or I need to look it up. One thing I want to explain is- and this is some magic which also like was a little bit weird to me in the first place, um I just like to find the right screen sharing. So one thing um like the documentation says you can include templates and you can like include built-in templates.

B

um The thing is: um is it linked over here yeah, it's linked already um so gitlab on uh on.com and also in the self-hosted version, provides templates by default, and this is shipped within the source code and being made available, for example, from the editor, so the golan template which we've been using is one of that, um but there are also others like the security testing security scanning and, for example, other devops uses the same mechanism in the background, and the interesting thing is- and I have this oh sorry going back over here- I wanted to copy it into this into the chat, so you can follow along um oops.

B

This basically puts us into the overview of the templates and, as you can see, there are quite a lot of them. One thing which we have been using is the go template.

B

Which is just over here, so you can either, for example, inspect it over here and reuse it or copy certain parts, or just like use the the official way of including a template.

B

um The other interesting thing is um there is, for example, the security sub folder, and this, for example, includes all the secret scanning and security scanning and so on, and this is basically where the magic happens, so it uses um a specific docker image which we provide. um There are certain techniques in in yammer or in the ci configuration which is like this is a template or like a skeleton provider, and you can use extends to import that into a job description.

B

um If you're going to look into that, let me see three yummy armor tips. This was the blog post. um This is something where this is described a little bit more in detail. How to use it in practice. um Just go over here.

B

Yeah can try it out and say hey. I don't want to write it all the time, um for example, defining the image or define this script. um You can let the jobs extend the template.

B

So it's a chop template in a specific sense. Other things which are interesting and to have a look at is, for example, the rules language which can get quite overwhelming. um This is the way to control and say hey. um This job should only be executed um when the default branch matches the commit branch, for example.

B

um Is it the other way around? You can control, for example, saying hey when this is a much request. I want to execute a specific job which does some more checks and when it's like, uh when the pipeline or the job is triggered in the main or master branch, um you, for example, trigger an automated deployment to production. um If you want to, for example, manage that in the network, um so these are basically the uh the ci templates and they are also developed over time.

B

So if you really want to like freeze it in your own environment, say hey, I don't want to always use what's in development over there, um you can um just like, can download the file or copy the file content in your own ci template repository and manage it, manage it over there and then, for example, you use the possibility to watch document and to have the documentation open.

B

I think it was this history.

B

You have the possibility to not only use include template but, for example, to use include remote, which means you have a public repository, and you just say remote in this url and fetch that from your central ci template repository, there is a way to include custom templates in your own installation. But to my knowledge, that's in the enterprise version.

B

It should be linked somewhere.

B

Custom templates.

B

Custom level templates the project templates now project templates or something else um too many features, um often times in terms of freezing.

B

If you really want to control what's going on, I would recommend to establish a review process um similarized, as you would develop code or source code and saying hey the this is the central repository for our ci templates and everyone is forced to use them and not just what gitlab provides um and if, for example, there is a new release of gitlab, um we are doing a diff or an automated div uh from from a job, um and you get like a report and saying: oh I'm acknowledging these changes or I'm like rejecting them, and I don't want to use them.

B

This is something you can do that help. Hopefully.

G

Yes, thank you.

A

Excellent, thank you. So much. We've got uh one more question, two more questions right now, uh one from flavio marquis and is ml ops supported. Somehow I will give flavio. I will give flavio uh the there we go. You you can you can ask your question directly. Flavio.

C

Okay uh thanks: uh well, the question is pretty clear: just autocuriosity as a for devops, even for the machine learning, slash ops, there is some support and in what extent.

B

um It's a good question. That's to be honest, it's not not my area, I'm still learning learning in that regard um is there. Is there a specific framework, you're looking into machine learning or like or you're just like starving.

C

Oh, no, nothing specific it's just uh to see because basically we are bringing gitlab in our uh corporation and uh we have a pretty big machine learning team and they will be really keen to eventually, you know, exploit the the the guitar functionality for the ops part for the operational part and yeah. I was just trying to to figure out for my side. You know how to answer to this kind of um expectation.

B

uh Then I'd love to give a shout out to my amazing teammate, renee hufflecht, override in the chat who answered that for me right now. um This is a blog post, which was recently published around machine learning and nci cd. um I would recommend checking this one out um and trying it out um looks interesting.

B

um To be honest, I haven't, haven't checked it out. Yet um the thing is.

B

B

It's amazing, when I just have to click links, um data science.

B

There's a video, oh interesting. um I know we're not watching the video now.

A

Precisely renee is uh raising his hand, so I will give him.

B

Oh yeah, that's even that's even better.

H

Yeah, thank you. um It's just um I'm not I'm not going to be able to be very in-depth here, but um I'm tending to enterprise customers in in this region and just in the last couple of months it the whole machine learning, data, mining, business intelligence, these topics, they they came up more and more in in communications, and it seems like that in in other areas of the world. um Other colleagues are further on. I know that in in asia, pacific there was a end of last year we had a.

H

We had a session, a special special team taking part in a machine learning event. So so it's definitely a topic, but currently it is not something that we have so much here, at least not so much knowledge in depth. Side of, but as you can see, it is definitely one of the uh one of the use cases um where github is also used.

H

Would that is that? Is it something? What what kind of what kind of things are you looking for? If you say mlops is what you have seen or what I posted going in the direction or is that something completely different.

C

uh No actually right in this direction. uh I cannot because I'm not you know a data scientist myself more on the system engineering side um but, as I said, no as we are bringing gitlab up to speed within our corporation for the devops part- and we are a very relatively huge for our corporation and a machine learning team that is growing and I had already collected a different feedback from them. They're really keen to to see the potentiality.

C

You know for even gitlab, probably because they're more self-aware than me, that no, that can be the case, as you mentioned, maybe from the other side of the world and more active on that. But we we're just experiencing not to bring the devops where gitlab uh in our system, engineering team, especially I work in splitter security.

C

uh But then we have this machine learning team. uh That is even keen to to to experience. That's why my my question was not really specific on some tech stack from the machine on this side, because not my my competence but yet yeah. I was just curious to see how eventually we can open this solution and github on prem. Even for uh for our machine learning team.

B

One one thing I could think of, and I'm I'm not really sure if we get there in 2021, have having sort of machine learning or code intelligence, you make mistakes often and then machine learning detects that pattern, and this kind of automatically creates a merge request, fixing the problems or some other machine learning. I'm I'm dreaming of. Is you want to have certain application, performance, monitoring or ci cd tracing, so you have some spans and traces.

B

You see which request takes long in your application and adding those measurements and points in your code as a developer is like, if you're not trained to that um it's it's cumbersome. It's like it's, it's uh it's not efficient and if we kind of get some like kraken or machine learning around this and saying hey um the robots or the bot does automatically add certain instrumenting to the code. This would totally be awesome.

B

The problem or the the thing is, I don't know how to make this happen, but I would love to have more automation in our code to make it even easier to, for example, monitor it or instrument it or just like prevent bad design patterns or bad code patterns from happening, because there is a bot out there which knows about this pattern.

A

Great thanks thanks a lot for the interventions, and we have one more question from mohamed havas. Is there a way to direct the computing of the build unit, test, test etc to a self-hosted cluster? I'm going to give mohamed again the possibility to talk to muhammad feel free to ask or to expand your question. If you want.

D

Hi, um sorry, you hear me yes,.

A

We hear you yeah.

D

Okay, cool- um I was wondering like, is there a way to use gitlab to basically just the.

D

Project and um basically like take the responsibility of computing, or we have like a whatever kind of like a repetitive task that I have to do for myself. Is there a gateway from gitlab to like a my own environment? Let's say.

B

So you mean that you're using gitlab.com um and add your own runners and your own clusters, or just everything self-hosted.

D

Yeah I just want to like use gitlab just like as a source control, but nothing more. I don't want to use.

D

I don't want to use it for continuous integration, but you know what like just like a general question like is there? Is there a way to um extend git lab, let's speak to to just like ship the jobs onto my setup.

B

So you mean that you, when.

B

You want to trigger an external ci cd system, so to speak, that what you're looking for.

D

So you head off.

B

B

You um I mean something around web hooks, um so when a specific event happens, um that you, for example, trigger an external tactile pipeline um or something else.

D

B

B

um I've seen something in that regard.

B

Well, it was a cute. Was it the cube can talk. I I think it was the cube contact.

B

I support text on pipeline. This is something else.

B

Can get the web box be used for triggering? I think this is one of the ways.

D

B

You could also trigger, for example,.

B

If you have your gitlab repository and you want to trigger github actions, this should be doable via web books and vice versa. You have a github repository and you want to make use of gitlab ci cd, for example.

B

This is also something that web hooks play play a big part in this integration um and also the rest api, a text on trigger okay, I think this is. This is the same one.

B

um I think it's doable, but um I think there is another single click application which does that for you right now, um another another way I could think of is, for example, using gitlab ci cd, um but actually triggering um the pipeline from the executed job like using the script, a script, keyword and then saying hey. I want to remotely trigger the tecton api.

B

Wait for the job do something around that another way, maybe writing a custom executor for the gitlab runner, which provides the different states which are which are then required, and this is done for the um is it the aws container, the farget integration, I think um custom custom executor, but this is a pretty much advanced thing to do um so.

B

The system is, it's meant to be extendable in a way of using different executors for the jobs, then um maybe it's an entry point to incorporate it more deeply with um with techtone, but I'm not not sure about this. This is just like experimental thinking right now,.

B

Probably the web hooks are the first step to look into.

D

Okay, thank you very much.

A

Thanks mohamed thanks michael and we have a raised hand by dan, I will give him the then you can you can see. Can you hear me? Yes, we can hear you.

E

All good, uh no, I wanted just to add to the question I was previously mentioned. Basically, you could watch for the webhook events when somebody pushes code to the repository and triggered ci from there. It's just you have to build your own watcher. So if you're trying to use gitlab and to execute something on push most likely, you would have to use the ci like gitlab, see every gitlab ci yaml.

E

We used to have.

A

E

Build onto google cloud build or if you want to really build your own.

A

E

The best because they do expose everything.

A

Unfortunately, we lost half of what you said.

E

I'm sorry so to push events. You basically have to build your own watcher since gitlab exposes a lot of events, so you can just listen to any type of different code event so such as a push, and if you really want to trigger that from your own pipeline, we used to have gitlab trigger on the free and shared runners, a google cloud build on our runners on google cloud, so that was an option. So if you have free credits on google cloud, it works really well.

D

Okay, I get it, I get it great.

A

Thanks a lot for that.

D

Thank you. Thank you very much. Notification.

A

D

A

And there's more comments on the chat as well. Rene as well has been commenting during this conversation, so there might be some interesting points as well there, so um we still have a little bit of time. I don't know if somebody would have would like to have a one last question, um some item that might be interesting for you for your organization, for your team take the opportunity the brains are here. You can ask them questions.

A

B

Yeah, maybe maybe I was talking too fast and um provided, uh triggered an overflow- I'm sorry about that when I'm, when I'm passionate, um I'm just like talking.

A

Now, but that's okay, that's okay! It was very good content and the good thing is that we have the both the the code and the the slides available. They are in the chat, so please everybody make sure that you download a copy of them and if there are no more questions I would say. Thank you so much michael, it's been a really really fantastic.

A

First meet up for this github gitlab group and yeah thanks again and thanks everyone for participating. I think it was fantastic. The the interventions, the questions, uh everybody with their own approaches and problems, and I think, that's very, very, uh very, very nice to hear how people are using gitlab in their own environment.

A

We are going to uh I'm going to quickly close this by telling you that please stay tuned on our vision.tv channel. We are going to be publishing this video in the next few days and on the meetup group. We are going to publish the date for the second edition of this meetup and again thanks a lot for your participation and see you soon in our next meetup group. Thank you so much bye. Everyone thanks guys.

E