This document describes how to run ESP with Google Cloud Endpoints integration on a Kubernetes cluster that can run anywhere as long as it has internet access.
-
Select or create a Cloud Platform Console project.
-
Enable billing for your project.
-
Note the project ID, because you'll need it later.
-
Install cURL for testing purposes.
-
Enable Cloud Endpoints API for your project in the Google Cloud Endpoints page in the API Manager. Ignore any prompt to create credentials.
To configure Endpoints, replace YOUR-PROJECT-ID with your own project ID in the swagger.yaml configuration file:
swagger: "2.0"
info:
description: "A simple Google Cloud Endpoints API example."
title: "Endpoints Example"
version: "1.0.0"
host: "YOUR-PROJECT-ID.appspot.com"
To deploy the sample application:
-
Invoke the following command:
gcloud service-management deploy swagger.yaml
The command returns several lines of information, including a line similar to the following:
Service Configuration with version "2016-04-27R2" uploaded for service "YOUR-PROJECT-ID.appspot.com"
Note that the version number that displayed will change when you deploy a new version of the API.
-
Make a note of the service name and the service version because you'll need them later when you configure the container cluster for the API.
To deploy to the cluster:
-
Edit the Kubernetes configuration file esp_echo_http.yaml, replacing SERVICE_NAME and SERVICE_VERSION shown in the snippet below with the values returned when you deployed the API:
containers: - name: esp image: b.gcr.io/endpoints/endpoints-runtime:0.3 command: [ "/usr/sbin/start_esp.py", "-p", "8080", "-a", "127.0.0.1:8081", "-s", "SERVICE_NAME", "-v", "SERVICE_VERSION", "-k", "/etc/nginx/creds/service-account-creds.json", ]
Note you also need to change the service type from LoadBalancer to NodePort if you use MiniKube
-
Create your service account credentials from Google API Console.
- Save your credential as
service-account-creds.json
- Save your credential as
-
Deploy the service account credentials to the cluster:
kubectl create secret generic service-account-creds --from-file=service-account-creds.json
-
Start the service using the kubectl create command:
kubectl create -f esp_echo_http.yaml
It can take a few minutes after you start your service in the container before the external IP address is ready.
To view the service's external IP address:
-
Invoke the command:
kubectl get service
-
Note the value for EXTERNAL-IP; you'll need it to send requests to the API.
After the sample API is running in the container cluster, you can send requests to the API.
To send a request to the API
-
Create an API key in the API credentials page.
-
Click Create credentials, then select API key > Server key, then click Create.
-
Copy the key, then paste it into the following export statement:
export ENDPOINTS_KEY=AIza...
-
-
Send an HTTP request using curl, as follows,
-
If you don't use Minikube:
curl -d '{"message":"hello world"}' -H "content-type:application/json" http://[EXTERNAL-IP]/echo?key=${ENDPOINTS_KEY}
-
Otherwise:
NODE_PORT=`kubectl get service esp-echo --output='jsonpath="{.spec.ports[0]}"'` MINIKUBE_IP=`minikube ip` curl -d '{"message":"hello world"}' -H "content-type:application/json" ${MINIKUBE_IP}:${NODE_PORT}/echo?key=${ENDPOINTS_KEY}
-
-
echo sample code (swagger.yaml in this directory is a copy from this echo sample)