This tool deploys an application to a Kubernetes Engine cluster, following Google's recommended best practices.

Using gke-deploy to deploy an application to Kubernetes Engine differs from kubectl in that gke-deploy is a wrapper around a kubectl apply deployment that follows Google's recommended best practices by...

• Modifying a set of Kubernetes configuration files to use a container image's digest instead of a tag

• Adding several recommended labels to the set of Kubernetes configuration files

• Getting authorized to access a GKE cluster

• Applying the set of Kubernetes configuration files that were modified in the prepare step, using kubectl apply

• Waiting for applied Kubernetes configuration to be ready

gke-deploy run [flags]

This command executes all of the best practices mentioned above, in sequence.

gke-deploy prepare [flags]

This command executes the steps to modify and add labels to a set of Kubernetes configuration files, and skips the steps to get authorized to access a GKE cluster, apply configuration, and wait.

gke-deploy apply [flags]

This command skips the steps to modify and add labels to a set of Kubernetes configuration files, and executes the steps to get authorized to access a GKE cluster, apply configuration, and wait.

View this page for examples on how to use gke-deploy with Cloud Build.

Follow these instructions to set up continuous deployment.

You can deploy an Application Custom Resource for your application:

1. Follow the instructions here to create or update a GKE cluster to install Application Delivery, which supports the Application CRD.

   You can also install the CRD manually by following the instructions here.

2. Use the --create-application-cr flag with gke-deploy prepare or gke-deploy apply to create an Application CR for your application.

Although gke-deploy is meant to be used as a build step with Cloud Build, you can run it locally for testing.

1. First, install kubectl as a dependency.

2. Next, install gke-deploy:

   go install github.com/GoogleCloudPlatform/cloud-builders/gke-deploy@latest
   gke-deploy -h

3. If your kubectl is pointing to a cluster, you can test out a deployment by deploying an application with one simple command:

   # Deploy an nginx Deployment with a load balancer exposing port 80.
   gke-deploy run -i nginx -x 80

4. After the command finishes successfully, gke-deploy will print a table displaying deployed resources. You can visit the IP address printed in the Service row.

   e.g.,

   

   curl 35.196.67.253