Skip to main content

Continuous Kubernetes deployments with Skaffold


I have been working with Docker containers and Kubernetes for quite some time now. When working with these technologies the usual workflow involves following steps (bare minimum)

  • containerize you application into a docker image
  • push the changes to the container registry like Docker Hub
  • deploy the containers to a cluster using container orchestration tool like Kubernetes

While developing an application, the developer also tests the application. It is quite time consuming to repeat the whole process every time there a change in the code and you need to rebuild the docker container, push it to registry and do the deployment. What if there was a tool which can do all of this for us and we focus on producing the code?

Introducing Skaffold

My dear friend who is a rockstar when it comes to introducing me to new cool tech, Baltazar Chua mentioned to me about Skaffold. Because Bal has this knack of picking the right tool for the right job, I have a hash tag for him #BalIsARockStar.

Skaffold is a command line tool that facilitates continuous deployment for Kubernetes applications.

In  this post we are going to do the following

  • Install Skaffold on Mac using Homebrew
  • Integrate Skaffold with existing multi container application
  • Test Skaffold using Minikube as target deployment environment

Install Skaffold on Mac using Homebrew

On the Skaffold GitHub repo, in the installation sections you find the steps to install on Linux or Mac OS. For Mac you run the following command

curl -Lo skaffold && chmod +x skaffold && sudo mv skaffold /usr/local/bin

I am a big fan of using Homebrew for installing packages on Mac. Luckily there is a Homebrew package available for Skaffold. I used this approach to install Skaffold on my Mac. Run the following command to get Skaffold installed.

brew install skaffold

Integrate Skaffold with existing multi container application

I will use the codebase from GitHub repo which I am currently working with the Azure Kubernetes Service (AKS) Learning series. This consists of a Web frontend, a web API project and a database running inside SQL Server 2017 Linux container.

Skaffold works on the basis of very little minimal configuration. First of all we need to define the configuration in a yaml file named skaffold.yaml. This file contains 3 phases.

1. Build

In the build phase we define the build artifacts which are docker images. In this step we specify the location of the Docker files that we need to build whenever there is change in the source code.

2. Push

Once the images are built, they are pushed to the container registry referenced in the image name. I am publishing the images to Docker Hub registry. When we use the Kubernetes cluster locally like minikube or Docker for Mac, the push is skipped because the images are available locally.

3. Deploy

The deploy step ensures that the most recent versions of the artifacts are running inside the cluster. I will be using kubectl to deploy. I can reuse the Manifest files that I had created earlier for the demos.

We can see the contents of the skaffold.yaml file

I have defined the build phase on line 3. On line 4 we list all artifacts to be built as part of this phase. Lines 5 to 10 specify the image names and the location of the Dockerfile.

Next is the deploy phase on line 11. We specify kubectl as a mode of deployment and finally the list of manifests are specified on line 13 & 14. I am using the wildcard option to deploy all the manifests under k8s/Minikube directory. Currently there are 8 different manifests as shown below.

lisf of manifests

With the configuration done, lets put the Skaffold to test.

Test Skaffold using Minikube as target deployment environment

We need t run the Skaffold dev command in the directory where the skaffold.yaml file resides. Check the output of the command

In this case you can see that the build was completed in less than 2 seconds and deployment completed in less than 5 seconds. This is a great beginning. For the first time, it can take a bit long to download the respective images and  build them locally. Once the images are available in the cache it is super fast.

Update source code and experience the magic

Lets test Skaffold by making a code change to the source code. In the screenshot below you see me commenting the block of code which shows the corousel control on the homepage. The moment I comment out the code, skaffold triggers a build and deploy. Its so fast that I could not capture it in full screen mode between the VS Code IDE changes, the iTerm terminal showing the automatic build triggering and then the Safari refreshing the final output. That is the reason, I kept the 3 windows open side by side so you can see in real time how the  changes are impacted. I did these commenting and uncommenting couple of times.

It would take minutes to redeploy the changes if we were to do it without Skaffold. The best part I like about Skaffold is it is completely aware of only the parts which are changed. Notice carefully how it triggers the build for only 1 docker image and subsequently deploys only the impacted service. This is freaking cool.

Another interesting point I found out about Skaffold was its cleanup mechanism. After the testing is completed, I can stop the continuous monitoring and deployment part by stopping the dev command (CONTROL + C). Skaffold does a neat job of cleaning up the resources it created itself.

skaffold cleanup

Gotchas & Limitations

Skaffold currently works only with Linux & Mac. There is no Skaffold available for Windows OS.

Looing at the Github page for installation section, it gives an impression that there is only Linux & Mac support for Skaffold. But as rightly pointed out by Baltazar Chua, there is support for Windows OS as well. You can download the Windows executable from

Refer to the detailed documentation for more details.

If you are using Minikube as the target cluster, the Minikube has to be up and running. Skaffold will not start the Minikube instance automatically for you.


As we can see from the quick demonstrations in this post Skaffold is quite a handy utility to add to your toolkit if you are developing containerized application and deploying them to Kubernetes cluster. I can see myself using it during the development phase to improve the inner loop workflow. Based on my past experience I can say that it is a huge productivity boost and can easily save few hours a day. I was using docker compose to build multiple images together. With Skaffold taking care of building, pushing and deploying images continuously, I don't feel the need for docker compose for my workflow. In this post we saw how to use skaffold with minikube Kubernetes cluster.

In the future post I will also demonstrate how to use it with the built in Kubernetes cluster with Docker for Mac.

Until next time, Code with Passion and Strive for Excellence.


Popular Posts

Build custom SQL Server 2017 Linux Docker image

BackgroundIn the earlier series of posts on continuous deployment of multi-container apps, I had used two containers for ASP.Net Core MVC web site and ASP.Net Core Web API. This post is a sort of continuation. In the earlier version we were using in memory database to store the data. This approach is useful in a proof of concept or a quick demo application. In an enterprise scenario we would use some sort  of persistent data store to store the data. It is very easy to containerize the stateless services and web front ends. When we move towards stateful processing, things start to get interesting.Containers by their very nature are supposed to be ephemeral. This means that containers can be stopped and destroyed and a new one built and put in place with minimum setup and configuration. If we destroy the container and replace it with a new one what happens to the data stored inside the container? Can we use containers to run database related processes? How can the stateless containers c…

Docker Compose Tip : How to avoid SQL Server 2017 container exiting when running using docker compose

Problem starting SQL 2017 containerWhile working on recent blog post on Integrating SQL Server 2017 Linux with ASP.Net Core using Docker I cam across a strange behavior. I was able to run SQL Server 2017 Linux container image successfully as demonstrated here. I was able to initialize the database using the command docker run -it -p 1433:1433 \ --name sql2017 \ nileshgule/sqldbThis command runs the custom image in interactive mode based on the –i flag. We also pass the t flag to create a psudo tty terminal. The next step for me was to try the integration with Dotnet Core project. I started with updating the docker-compose file and other changes explained in the post on integration.Before I added the sql2017 as the service inside docker-compose file, I ran the SQL Server 2017 container individually using the above command & verified that the Core MVC and Core Web API we able to communicate to the database inside container.This seemed pretty easy at first. All I had to do was update…

Integrate SQL Server 2017 Linux with ASP.NET Core using Docker

In the previous post, we saw how to initialize a SQL Server 2017 on Linux database running inside Docker container. This post is continuation of my exploration about Docker containers for stateful applications. In this post we will explore how to integrate the SQ Server 2017 on Linux container with ASP.Net Core Web API also running inside a Docker container. We will perform following activitiesUpdate Docker compose file to build the SQL server imageUpdate Web API project to use SQL Server Database instead of In memory databaseBuild and run the application containers using Docker compose commandsQuery Data from SQL container using the MVC web front endUpdate Docker compose file to build the SQL server imageDocker provides Docker Compose as a tool for defining and running multi-container Docker applications. I already have an existing docker compose file which I have been using for my series of post on implementing continuous deployment of multi container apps. In a compose file we defi…