Skip to content

Example of techniques for using Google Cloud Build to do integration tests of a microservices application

License

Notifications You must be signed in to change notification settings

endocode/cloudbuild-integration-testing

 
 

Repository files navigation

Integration Testing on GCB

This is a demo showing how to execute multi-container integration tests as part of a Google Cloud Build invocation.

Download the latest stable release of this demo here.

Method 1: docker-compose

Prerequisites

Build the docker-compose community builder and push to Google Container Registry in your GCP project

Running Build

Run command:

gcloud builds submit --config=cloudbuild.compose.yaml .

Method 2: Google Kubernetes Engine

Overview

2a: Deploying to an existing kubernetes cluster: cloudbuild.gke.yaml

2b: Deploying to a new cluster per test: cloudbuild.gke-per-test.yaml

Prerequisites

  1. Create a cluster in Google Kubernetes Engine

    gcloud container clusters create staging --zone us-central1-c
    

    NOTE: Update cloudbuild.gke.yaml env options if using a cluster with a different name or zone.

  2. Allow traffic on default potential NodePort range

    gcloud compute firewall-rules create allow-k8s-nodeports --allow tcp:30000-32767
    
  3. Add Kubernetes Engine IAM role to Cloud Build Service Account

    Method 2a: Deploying to existing Kubernetes cluster:

    gcloud projects add-iam-policy-binding <PROJECT-ID> \ 
    --member serviceAccount:<PROJECT-NUMBER>@cloudbuild.gserviceaccount.com \
    --role roles/container.developer
    

    Method 2b: Deploying to a new cluster per test:

    gcloud projects add-iam-policy-binding <PROJECT-ID> \ 
    --member serviceAccount:<PROJECT-NUMBER>@cloudbuild.gserviceaccount.com \
    --role roles/container.admin
    

    Learn more about the Cloud Build Service Account, Kubernetes Engine Permissions and Granting Roles to Service Accounts.

Running Build

Deploying to an existing kubernetes cluster:

gcloud builds submit --config cloudbuild.gke.yaml .

Using a new kubernetes cluster per test:

gcloud builds submit --config cloudbuild.gke-per-test.yaml .

When you're done

  1. Delete Kubernetes Cluster
    gcloud container clusters delete staging --zone us-central1-c
    
  2. Remove GKE permissions from Cloud Build
    gcloud projects remove-iam-policy-binding <YOUR-PROJECT-ID> \ 
    --member serviceAccount:<YOUR-PROJECT-NUMBER>@cloudbuild.gserviceaccount.com \
    --role roles/container.<developer-or-admin>
    
  3. Remove firewall rule
    gcloud compute firewall-rules delete allow-k8s-nodeports
    

Method 3: Deploy to self-destructing VM

Before beginning, update k8s/db.yaml and k8s/web.yaml with your Project ID.

to do things locally:

# run terraform to create a self-destructing VM w/ microk8s
terraform apply -var="project-name=$(gcloud config get-value project 2> /dev/null)" -var="instance-name=test-$(date +%s)" -auto-approve

# get the IP from terraform
echo $(terraform output ip) > .tmp.microk8s_ip

# patch the IP into the kubectl config
# (backup flag is passed for macOS compatibility)
# TODO: use a better regex so this can work repeatedly instead of just once
sed -i.sed-bak "s/CLUSTER_IP/$(< .tmp.microk8s_ip)/" kubeconfig.microk8s

# kubectl can now deploy to microk8s
kubectl apply -f ./k8s --kubeconfig=kubeconfig.microk8s

# see services
kubectl get services --kubeconfig=kubeconfig.microk8s

to do things in Cloud Build:

(prerequisite: terraform builder is built and pushed to GCR in project)

gcloud builds submit --config=cloudbuild.vm.yaml .

This is not an official Google product.

About

Example of techniques for using Google Cloud Build to do integration tests of a microservices application

Resources

License

Code of conduct

Stars

Watchers

Forks

Packages

No packages published

Languages

  • JavaScript 55.2%
  • Shell 21.4%
  • HCL 15.1%
  • CSS 4.2%
  • Pug 2.2%
  • Dockerfile 1.9%