Cloud Native Computing Foundation KubeCon + CloudNativeCon North America 2018 (Seattle), 15 Dec 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Lightning Talk: Spawning Kubernetes In CI For Integration Tests - Marko Mudrinić, Loodse

Description

Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io

Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects

Lightning Talk: Spawning Kubernetes In CI For Integration Tests - Marko Mudrinić, Loodse

Making sure a Kubernetes operator works correctly is a very important step before deploying it to production. Without good integration and E2E test coverage, it’s easy for a bug to get in.Therefore everybody who is seriously developing operators will face the problem of spawning Kubernetes in CI: integration tests assume a complete Kubernetes cluster. There are many solutions for bootstrapping clusters, but many of them don’t work in CI or require additional configuration.In this talk we’ll take a quick look at the most popular solutions for running Kubernetes in CI, such as Minikube and DIND. We’ll compare those solutions and see what the requirements for each of them are.

To learn more: https://sched.co/GrUv

A

Hello, everyone I am Marco, and today we are going to talk a little bit about running kubernetes in CI, so we often heard that integration tester import a part of every project and story date. Project integrates well with other components: humanities controllers and operators are no different. We want to be sure that they work when deployed in an actual cluster, but of course for that we need a real Cuban at this cluster, and the question is how we can get one running in our CI pipeline.

A

There are many solution for running kubernetes such as mini cube, cube, ADM DND, but many solution requires systemd, use it to start and manage all services needed for a cluster. The problem means that CI environment is usually minimal or an older operating system version such as Ubuntu 4004, with no access to system D, making it impossible to use such solutions.

A

Today we are going to talk little bit about kind. It is a new project by Q Benitez community, using for running local kubernetes clusters using just docker. He supports cuban. It is 1.11 inure. So let's see why kind? The only requirement for any kind is dr. of a label on the machine. Kind comes with a prebuilt docker image called base image, which has all dependencies pre-configured, including system d, docker and cube ADM.

A

When you create a new kind cluster, it starts a docker container using the base image and then, in that base image creates a single node cuban I discussed using hubei DM, the usage of cube ADM and the cost for base images make. It is a customizable to start caster. However, you want other using flags or by using customize it base image.

A

So we are going to see how that looks like Forte VCI and how we can get kind caster running there. The VCI is of the most popular CI systems among the open-source project, but these steps should work for any other CI system as long as you have a label. So this is one manifest we're at the top of it. We define what the vironment we use with go and docker.

A

Then we define job which first double scoop cuddle. This is way to do that is to grab it from the cue cuddle bucket and then just move it to our pet. Then we don't kind with this go. Get comment kind, also have binary some github realizes a surprisedly, so you can use the similar step as for cue cuddle, if you're right running the latest versions and more stable, and then we create a new cluster using kind create caustic comment.

A

This comment takes a few minutes to set up everything and after it is done, everything we need to do is to export cube, coughing environment variable and then our cluster is ready to be used in cube. Kado can interact with our cluster, then. The other thing we have to do is to run integration tests and you can run whatever script you want here and, for example, we use some a comment here, but there is one more problem here and this is about building. The document is to be used by kind clusters.

A

So when you deploy some Tegel kubernetes clusters, it either download the image from a documentary or some other registry or uses the local image on the cluster. That is already available so because, as mentioned earlier, we have the concept: where kind is running a container where humanity is being run, we need to pass our image to that class. To that container option one is to run container registry and to point kind to use that try to say, but that of the camp is some little bit harder.

A

So another option is to cite all the image to docker running in kinds container. So there is another feature for that. That is working progress and there is issue 28 on code repository that you can follow the progress, but in meanwhile there is a sample script for side loading available in the J stock search manager, fo that you can use for your project to build images that are available in height, and this is how sub script looks like SP, but you have time to go through it.

A

It will be available in the repository or do this talk so I can left a repository, X, mu, V, Davis kind. You can also find me on Twitter it's like if you have questions, but there is also a kind channel on communities like hold kind. If you have any other questions, thank you for coming today. Thank.

B

You Marco right on time, just a minute over and that was Marcos. First talk ever at the conference. He told me so give him another round of applause. It's not easy! Standing in front of a group.