jetty-managed-vm

Cloud Bigtable on Managed VM's
(Hello World for Cloud Bigtable)

With the Jetty runtime one can debug in the Docker container on a local machine.

This app provides:

1. A web interface that uses Cloud Bigtable to track the number of visits from an opaque version of your Google account.
2. A simple REST interface that can read and write arbitrary data to a Cloud Bigtable table using GET, POST, and DELETE verbs.

SECURITY WARNING - This app provides NO SECURITY protections for the two tables you create (gae-hello and from-json) are open to any user on the internet through this app. We suggest that instances should only be available while testing and that test data be used.

Table of Contents

1. Requirements
2. Docker on a Mac
3. Project Setup
4. Using Jetty Runtime Locally
5. Deploying the Jetty Runtime
6. Using Bigtable-Hello
7. Using-JSON

Requirements

1. Latest version of gcloud Update with gcloud components update
2. Docker
3. Java 1.7

Docker on a Mac

All machines on the internet have a preset DNS entry known at localhost which maps to an IP address of 127.0.0.1. Accessing local services, can usually be done by going in your browser to localhost:8080. Docker runs inside a VM on your Mac, that VM has it's own IP Address which can be found using boot2docker ip. (Typically this is 192.168.59.103). The sample uses Google Sign-in to create a unique id for each user. Google Sign-in requires that all hosts be accessed by name. So, on a Mac, it is necessary to modify your /etc/hosts (often done by sudo vi /etc/hosts - if you know how to use vi) file to add in an entry for docker. It should look like:

127.0.0.1	    localhost
255.255.255.255	broadcasthost
::1             localhost
192.168.59.103  docker

• If boot2docker isn't already running, start it: boot2docker start
• If you ever change IP addresses, you'll need to boot2docker restart

Project Setup

1. Follow the instructions for Creating a Google Developers Console project and client ID

Please be sure to add http://docker:8080 (if you are on a Mac), http://localhost:8080, and https://projectID.appspot.com as Authorized Javascript Origins

1. Use Cloud Console to enable billing.

2. Select APIs & Auth > APIs

3. Enable the Cloud Bigtable API and the Cloud Bigtable Admin API
   (You may need to search for the API.)

4. Select APIs & Auth > Credentials

5. Select Generate new JSON key

6. Set the environment variable GOOGLE_APPLICATION_CREDENTIALS to point to your json key

Many users find it helpful to add to either their .bash_rc or .profile the line:
export GOOGLE_APPLICATION_CREDENTIALS=~/path_to_key.json

1. Select Storage > Cloud Bigtable > New Cluster

Create a new Cluster -- You will need both the Zone and the Unique ID

1. Using gcloud, login.

gcloud auth login

1. Follow the instructions to enable hbase shell

2. Launch hbase shell

3. Create the table (tableName, Column Family)

create 'gae-hello', 'visits'
create 'from-json', 'cf1', 'cf2', 'cf3', 'cf4' exit

Using Jetty Runtime Locally

This describes a Jetty based Servlet that has been made into a Custom Runtime for Google Managed VMs -- This means that you do not have access to the normal AppEngine API's.

1. Edit Dockerfile to set BIGTABLE_PROJECT, BIGTABLE_CLUSTER, and BIGTABLE_ZONE (if necessary)

2. Edit src/main/webapp/index.html to set google-signin-client_id

3. Copy your keyfile *.json to src/main/webapp/WEB-INF

4. Edit Dockerfile, uncomment and modify the line (if you will be running locally)

env GOOGLE_APPLICATION_CREDENTIALS=/app/WEB-INF/YOUR_KEY_FILE.json

1. Build the java artifacts and docker image

   mvn clean compile process-resources war:exploded && docker build -t bigtable-hello .
   

2. run the docker image

   docker run -p 8080:8080 bigtable-hello

3. go to docker:8080 (Mac) or localhost:8080 (Linux) in your browser.

Deploying the Jetty Runtime

(First build & Run Locally)

1. If you haven't already done so, set your project in the gcloud tool.

gcloud config set project PROJECT_ID

1. Deploy the application

gcloud preview app deploy app.yaml --docker-build=remote

1. go to the new default module which will be displayed in results from the deploy. It will look like: https://20150624t111224-dot-default-dot-PROJECTID.appspot.com

NOTE - This is not the default version - which Google Auth requires (well, you could specify a version in the list of Authorized referrers, but that would be long winded), so, you need to visit the cloud console Compute > App Engine > Versions and make your version the default. Then test using https://projectID.appspot.com. If your test works, then you'll want to delete old versions eventually.

Using Bigtable-Hello

1. With your browser, go to docker:8080 (Mac) or localhost:8080 (Linux) in your browser. (Local) Or to https://.appspot.com

2. Sign-in with Google. Afterwards, your visit should increment the counter.

Using JSON

1. Entities (rows) can be accessed using //projectID.appspot.com/json/rowkey

• GET - Will wrap up everything as JSON
• POST - Will convert the JSON to ColumnFamily : Qualifier and write the data
• DELETE - Will remove the row.

1. The URL should be either localhost:8080, docker:8080, or https://.appspot.com

2. curl -H "Content-Type: application/json" -X POST -d '{"username":"red","id":535}' http://localhost:8080/json/blueword

3. curl -X GET http://localhost:8080/json/blueword

4. curl -H "Content-Type: application/json" -X DELETE http://localhost:8080/json/blueword

You will note that none of these examples use Column Family specifiers. It defaults to using cf1, if you wish to use the other column families specify <columnFamily>:<column> where columnFamily is one of cf1, cf2, cf3, or cf4 that you created earlier.