Managing software infrastructure is always a challenge. Kubernetes (also known as K8s) is an excellent platform that leverages containers to manage all stages of your project pipeline. It was originally an internal project at Google called Borg before it was made open source. Today, developers around the world use K8s with the backing of the open source community.
We’ll cover some tools you can use with K8s to both build and manage your infrastructure. But first, let’s define what Kubernetes actually is and quickly review the following topics.
Since many of these tools require having an existing cluster to work with, we’ll walk through a basic setup using the Google Cloud Kubernetes Engine (GKE) as well.
All code that is used in this post can be found in my GitHub repo.
Kubernetes manages applications that are deployed in containers. Containers provide mobility and fine-grained control over applications in all stages of the product lifecycle.
K8s itself operates with a control plane and worker nodes, as you see in the following diagram.
(Source: Kubernetes official documentation)
The elements within the control plane do the work of managing the infrastructure itself. This is where the actual management occurs. Using the various tools I’m about to describe, you pass commands to the different components within the control plane to apply changes and functionality to the nodes.
All of this together forms a K8s cluster. Typically, you’ll have a master node that has the elements in the control plane, and then you’ll have worker nodes where the actual application containers run.
Two other terms you often hear when working with Kubernetes are deployment and service. Deployments refer to configurations that make up a cluster. This is normally in the form of a config YAML file, but there are other ways to create deployments using docker images and other resources. Services refer to an abstract representation of an application running in a container within a node.
To really see K8s tools in action, it helps to have a working cluster that you can interact with. In my GitHub repo, I’ve included both instructions for building a cluster with Google Cloud Kubernetes Engine and Minikube.
Once you have these set up, you can use these examples to test the tools I’ll cover in this article. It also helps to have kubectl already installed.
kubectl enables you to:
If you want to first create a cluster, you can apply a deployment via a YAML file, like this:
> kubectl apply -f deployment.yaml deployment.extensions/helloworld-gke created
Once you’ve created a deployment, you can get the status of the clusters running.
> kubectl get deployments NAME READY UP-TO-DATE AVAILABLE AGE helloworld-gke 1/1 1 1 11s
Below are some other tasks you can accomplish with
Get information about the pods in a cluster:
➜ google-cloud git:(master) ✗ kubectl get pods NAME READY STATUS RESTARTS AGE helloworld-gke2-554f48b47b-69lbc 1/1 Running 0 6m5s ➜ google-cloud git:(master) ✗
Create a service via a config file:
> kubectl apply -f service.yaml service/hello created
Get information about a service:
> kubectl get services NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE hello LoadBalancer 10.31.247.92 <pending> 80:32294/TCP 31s kubernetes ClusterIP 10.31.240.1 <none> 443/TCP 122m
View logs within one of your pods:
➜ google-cloud git:(master) ✗ kubectl logs helloworld-gke2-554f48b47b-69lbc > [email protected] start /usr/src/app > node index.js Hello world listening on port 8080 Hello world received a request.
There are a lot more options with kubectl. For more, check out the K8s cheat sheet.
While kubectl enables you to interact with your cluster as a whole,
kubefed enables you to interact with a cluster via a control pane.
As I stated earlier, a control plane is the part of K8s that manages a cluster’s worker nodes. In a larger application, you may have multiple clusters that need to be managed.
kubefed enables you to interact with the cluster (or clusters) from a higher level of federated control. This is particularly good when considering security options, such as setting up TLS for your clusters.
The following example command deploys a federation control plane with the name
fellowship, a host cluster context
rivendell, and the domain suffix
Some examples include adding a control plane with a cluster.
kubefed init fellowship \ --host-cluster-context=rivendell \ --dns-provider="google-clouddns" \ --dns-zone-name="example.com."
This example (copied from the K8s reference docs) deploys a federated control plane with the name of
fellowship and context of
host-cluster controls the rest of the clusters in your federated system.
It’s also possible to add clusters to a control plane. Once you’ve created a control plane, you can add a cluster with something like this:
kubectl create clusterrolebinding <your_user>-cluster-admin-binding --clusterrole=cluster-admin --user=<your_user>@example.org --context=<joining_cluster_context>
kubefed works with kubectl and is very powerful. Refer to the K8s docs for more info.
Often while working with K8s, you’ll want to test something out on an individual pod before you apply it to an entire cluster. Minikube is a tool that allows you to build a one-node cluster on your local machine. Here you can test out what your nodes would look like with various configuration changes. The advantage is that you can easily create containers without being concerned about impacting a larger cluster.
Setting up Minikube will depend on the hardware you are using. The steps below work for a Mac, but you can check the docs for a more detailed walkthrough.
The first step to set up Minikube is to verify that virtualization is available on your machine.
sysctl -a | grep -E --color 'machdep.cpu.features|VMX'
You should see something like this:
machdep.cpu.features: FPU VME DE PSE TSC MSR PAE MCE CX8 APIC SEP MTRR PGE MCA CMOV PAT PSE36 CLFSH DS ACPI MMX FXSR SSE SSE2 SS HTT TM PBE SSE3 PCLMULQDQ DTES64 MON DSCPL VMX SMX EST TM2 SSSE3 FMA CX16 TPR PDCM SSE4.1 SSE4.2 x2APIC MOVBE POPCNT AES PCID XSAVE OSXSAVE SEGLIM64 TSCTMR AVX1.0 RDRAND F16C
Next, you’ll want to install it with brew.
brew install minikube
Confirm installation with the following.
Once the console output finishes, you can verify that your cluster is working correctly with the
minikube status. You should see something similar to the following.
➜ ~ minikube status host: Running kubelet: Running apiserver: Running kubeconfig: Configured
Now that Minikube is installed, you can create a one-node cluster. You can either do this with images from a Docker registry or locally built images.
Note that Minikube runs entirely in a VM. When you
stop Minikube, you are basically shutting down the VM.
To create deployments against your locally running Minikube, you can either pull a Docker registry image or use the local Docker daemon in your Minikube VM.
eval $(minikube docker-env)
docker build -t helloworld-minikube
kubectl run helloworld-minikube --image=helloworld-minikube:latest --image-pull-policy=Never
kubectl expose deployment helloworld-minikube --type=NodePort --port=808
➜ minikube git:(master) ✗ minikube service helloworld-minikube --url http://192.168.64.6:32100 ➜ minikube git:(master) ✗ curl http://192.168.64.6:32100 Hello World from your local minikube!%
The cool part about this setup is that you can just use local images; you don’t have to actually push it into a registry.
Overall, the primary advantage of using Minikube is that you can experiment and learn without worrying about the confines of a larger system. For more information, check out my GitHub project and the K8s docs on Minikube.
When working with K8s, it’s helpful to have a single source of information on your cluster. Dashboard is a web interface that lets you monitor the state of your cluster, and it can be run both locally and in deployed environments. You can run the Dashboard in your hosted instances as well as a local setup such as Minikube. Dashboard is a very nice way to quickly see the state of your cluster and its nodes.
To deploy the dashboard locally on top of your running clusters, just run the following with kubectl.
kubectl apply -f https://raw.githubusercontent.com/kubernetes/dashboard/v2.0.0-beta8/aio/deploy/recommended.yaml
Then, run the kubectl proxy.
Next, you can access the dashboard via the following URL on your local machine:
You should see the following output.
(Source: Kubernetes official documentation)
From here, you can access logs and many other features.
Developers within the K8s community are working on several additional tools beyond what we’ve covered here. I’ll briefly describe a few.
Helm allows you to manage packages used by your K8s infrastructure. These are called “charts,” and they enable you to abstract away package management. The nice part about Helm is that you can use preconfigured packages, or you can package up your existing applications.
If you’re familiar with Docker but not K8s, Kompose enables you to convert a Dockerfile into a K8s config file for deployments. There are a lot of cool things you can do with this.
Istio is a very popular open-source framework that is now available to manage message passing in your clusters. Istio does what a lot of the tools I’ve described in this post already do: if you set up Istio to work with your cluster, you’ll have a convenient third-party tool that can streamline communication between your cluster and nodes.
There are many other open-source projects that help with K8s. Typically, you see these in the form of either frameworks that can run in your control plane or SDKs that enable your containers to communicate with one another. The popularity and community behind K8s make working with this framework both fun and exciting.
I hope you’ve been able to learn something from the tools I’ve shown here. I highly recommend checking out the K8s GitHub repo to learn more about what the open source community is doing. There is a lot of cool things going on, and it’ll be exciting to see how Kubernetes evolves in 2020.
Install LogRocket via npm or script tag.
LogRocket.init() must be called client-side, not
CRDTs, or conflict-free replicated data types, is a concept that underlies applications facing the issue of data replication across a […]
We explore the fusion of TensorFlow and Rust, delving into how we can integrate these two technologies to build and train a neural network.
SignalDB enables automatic data synchronization between your components and a local in-memory or persistent database.