Modernize and Extend: JPetStore on IBM Cloud Kubernetes Service

This demo modernizes an existing Java web application (JPetStore) by:

  1. building Docker containers from the legacy stack
  2. moving the app to IBM Cloud Kubernetes Service
  3. and extending it with Watson Visual Recognition and Twilio text messaging (or a web chat interface).

IBMers can access the demo script and additional collateral from here.

Containerized Applications with IBM Cloud Kubernetes Service

Before you begin

Follow the below steps to create IBM Cloud services and resources used in this demo. You will create a Kubernetes cluster, an instance of Watson Visual Recognition and an optional Twilio account (if you want to shop for pets using text messaging).

  1. If you do not have Docker or Kubernetes tooling installed, see Setting up the IBM Cloud Developer Tools CLI.

  2. Set up a cluster by going to the Kubernetes Service on IBM Cloud and provision a Standard Paid cluster (it might take up to 15 minutes to provision, so be patient). A Free cluster will not work because this demo uses Ingress resources.

  3. Follow the instructions in the Access tab of your cluster to gain access to your cluster using kubectl.

  4. Visit IBM Cloud catalog and create a Watson Visual Recognition service (choose the Lite plan). After creation, you will get a set of auto-generated service credentials. The apikey and URL are required later.

  5. Optionally visit Twilio, sign up for a free account and buy a number with MMS capabilities by creating a project/feature on the Dashboard. Locate the Account SID and Auth Token from the API Credentials in the dashboard. Locate you Phone Number on the respective Twilio page.

Automated deployment

A toolchain has been created and automates deployment of the demo. You will still need to manually configure Twilio as described in the Manual deployment section.

Create toolchain

Once the toolchain has completed, the applications will be available at https://jpetstore.<your-cluster-ingress-domain> and https://mmssearch.<your-cluster-ingress-domain>.

The toolchain includes a stage named UNINSTALL (manual). This stage can only be triggered manually and will remove all resources created by the toolchain.

Manual deployment

To manually deploy the demo, follow the below steps.

Clone the demo to your laptop

Clone the demo repository:

git clone https://github.com/ibm-cloud/jpetstore-kubernetes
cd jpetstore-kubernetes

Code structure

Folder Description
jpetstore Traditional Java JPetStore application
mmssearch New Golang microservice (which uses Watson to identify the content of an image)
helm Helm charts for templated Kubernetes deployments
pet-images Pet images (which can be used for the demo)

Set up Watson Visual Recognition

  1. Create a file with the name mms-secrets.json by using the existing template:

    # from jpetstore-kubernetes directory
    cd mmssearch
    cp mms-secrets.json.template mms-secrets.json
  2. Run ibmcloud ks cluster get --cluster CLUSTER_NAME to get your Ingress Subdomain (make sure to replace CLUSTER_NAME with the name of the cluster you created above).

  3. Open mms-secrets.json file and update the Ingress Subdomain in the jpetstoreurl field. This allows the mmssearch microservice to find the images that are part of the message back to the user. Example: http://jpetstore.yourclustername.us-south.containers.appdomain.cloud

  4. Go to the IBM Cloud catalog and create a Watson Visual Recognition service (choose the Lite plan).

  5. After creation, you will get a set of auto-generated service credentials. Carefully copy the API key and URL into the watson section of mms-secrets.json file.

Set up Twilio (optional)

This step is only required if you want to use MMS text messaging during the demo (which is not possible in many countries outside the U.S.).

Skip this section if you only want to interact using the web chat.

  1. Visit Twilio, sign up for a free account and buy a number with MMS capabilities by creating a project/feature on the Dashboard.

  2. Open the mms-secrets.json file and replace:

    1. sid and token values with your AccountSID and the AuthToken from the Twilio Account Dashboard.
    2. number with the number you just purchased in the format +1XXXYYYZZZZ.
  3. Configure Twilio to send messages to the MMSSearch service

    1. Go to Manage Numbers on Twilio dashboard by clicking on All Products & Services on the left pane then click on your number.
    2. In the Messaging section of the Configure page under A message comes in, select Webhook, set the URL to http://mmssearch.<Ingress Subdomain>/sms/receive and the METHOD to HTTP POST

Create Kubernetes secrets

Next, use the kubectl command to allow your Kubernetes cluster access to the secrets you just created. This will allow it to access the visual recognition and Twilio services:

# from the jpetstore-kubernetes directory
cd mmssearch
kubectl create secret generic mms-secret --from-file=mms-secrets=./mms-secrets.json

Build and push the container images

The docker images for each of the micro-services need to be built and then pushed to a container registry. Here are the steps for pushing to your IBM Cloud private registry, but be aware that you could also push them to a public registry.

  1. Identify your registry Namespace with ibmcloud cr namespaces or create a new one using ibmcloud cr namespace-add <NAMESPACE>

  2. Set MYNAMESPACE env var to your namespace.

    export MYNAMESPACE=<NAMESPACE>

  3. Identify your Container Registry (e.g. us.icr.io) by running ibmcloud cr info.

  4. Set MYREGISTRY env var to your registry.

    export MYREGISTRY=<REGISTRY>

  5. Make sure that the steps above worked by running echo ${MYREGISTRY}/${MYNAMESPACE} . You should see output similar to us.icr.io/mynamespace

  6. Build and push the jpetstoreweb image. Run these commands as they are. You do not need to replace any of the values belwo:

    # from the jpetstore-kubernetes directory
    cd jpetstore
    docker build . -t ${MYREGISTRY}/${MYNAMESPACE}/jpetstoreweb
    docker push ${MYREGISTRY}/${MYNAMESPACE}/jpetstoreweb

    If you see Unauthorized while pushing the image, run ibmcloud cr login to ensure you are logged into the IBM Cloud and have access to the container registry.

  7. Next, build and push the jpetstoredb image:

    # from the jpetstore directory
    cd db
    docker build . -t ${MYREGISTRY}/${MYNAMESPACE}/jpetstoredb
    docker push ${MYREGISTRY}/${MYNAMESPACE}/jpetstoredb
  8. Build and push the mmssearch image:

    # from the db directory
    cd ../../mmssearch
    docker build . -t ${MYREGISTRY}/${MYNAMESPACE}/mmssearch
    docker push ${MYREGISTRY}/${MYNAMESPACE}/mmssearch
  9. Finally make sure that all three images have been successfully pushed to the IBM Cloud container registry by running ibmcloud cr images --restrict $MYNAMESPACE .

Deploy the application

There are two different ways to deploy the three micro-services to a Kubernetes cluster:

Option 1: Deploy with Helm (recommended)

  1. Install Helm. (brew install kubernetes-helm on MacOS)

  2. Find your Ingress Subdomain by running ibmcloud ks cluster get --cluster YOUR_CLUSTER_NAME , it will look similar to "mycluster.us-south.containers.appdomain.cloud".

  3. Open ../helm/modernpets/values.yaml and make the following changes.

    • Update repository and replace <REGISTRY> with your Container Registry and <NAMESPACE> with your Container Registry namespace.
    • Update hosts and replace <Ingress Subdomain> with your Ingress Subdomain.
  4. Repeat the previous step and update ../helm/mmssearch/values.yaml with the same changes.

  5. Next, install JPetStore and Visual Search using the helm yaml files you just created:

    # Change into the helm directory
    cd ../helm
    
    # Initialize helm
    helm init
    
    # Create the JPetstore app
    helm install --name jpetstore ./modernpets
    
    # Ceate the MMSSearch microservice
    helm install --name mmssearch ./mmssearch

Option 2: Deploy using YAML files

If you did not deploy using Helm, you can deploy using the yaml files and kubectl. For this option, you need to update the YAML files to point to your registry namespace and Kubernetes cluster Ingress subdomain:

  1. Edit jpetstore/jpetstore.yaml and jpetstore/jpetstore-watson.yaml and replace all instances of:

    • <CLUSTER DOMAIN> with your Ingress Subdomain (ibmcloud ks cluster get --cluster CLUSTER_NAME)
    • <REGISTRY NAMESPACE> with your Image registry URL. For example:us.icr.io/mynamespace
  2. kubectl create -f jpetstore.yaml - This creates the JPetstore app and database microservices

  3. kubectl create -f jpetstore-watson.yaml - This creates the MMSSearch microservice

You're Done!

You are now ready to use the UI to shop for a pet or query the store by sending it a picture of what you're looking at:

  1. Access the java jpetstore application web UI for JPetstore at http://jpetstore.<Ingress Subdomain>/shop/index.do

  2. Access the mmssearch app and start uploading images from pet-images directory.

  3. If you configured Twilio, send a picture of a pet to your Twilio number via your phone. The app should respond with an available pet from the store or or with a message that this type of pet is not available:

Using your Mac to send text messages to Twilio

If you'd like to send and receive texts from the pet store on your Mac, do the following steps:

  1. Ensure your iPhone is capable of forwarding text messages to your Mac.
    • See this Apple support document.
    • If the Text Message Forwarding option is not present, confirm that your Apple ID is enabled under Send & Receive.
  2. Access the Getting Started page from your Twilio account home page
  3. In the Send a Message widget, enter the Twilio number you bought into the To: text field.
  4. Add a message to the Body text field and click the Make Request button.
  5. After receiving the message on your Mac, drag and drop an image into the iMessage window

Logging

Check this tutorial - Analyze logs and monitor application health with LogDNA and Sysdig

Monitoring

Check this tutorial - Analyze logs and monitor application health with LogDNA and Sysdig

Load Generation for demo purposes

In a demo situation, you might want to generate load for your application (it will help illustrate the various features in the dashboard). This can be done through the loadtest package:

# Use npm to install loadtest
npm install -g loadtest

# Geneate increasing load (make sure to replace <Ingress Subdomain> with your ingress subdomain)
loadtest http://jpetstore.<Ingress Subdomain>/

Clean up

# Use "helm delete" to delete the two apps
helm delete jpetstore --purge
helm delete mmssearch --purge

# Delete the secrets stored in our cluster
kubectl delete secret mms-secret

# Remove the container images from the registry
ibmcloud cr image-rm ${MYREGISTRY}/${MYNAMESPACE}/mmssearch
ibmcloud cr image-rm ${MYREGISTRY}/${MYNAMESPACE}/jpetstoreweb
ibmcloud cr image-rm ${MYREGISTRY}/${MYNAMESPACE}/jpetstoredb

# Delete your entire cluster!
ibmcloud ks cluster rm --cluster yourclustername

Troubleshooting

The toolchain complains about "incompatible versions" for helm

The DEPLOY log shows:

Error: UPGRADE FAILED: incompatible versions client[v2.8.1] server[v2.4.2]

It means you have already helm installed in your cluster and this version is not compatible with the one from the toolchain.

Two options:

  1. Before running the toolchain again, update helm to the version used by the toolchain on your local machine then issue a helm init --upgrade against the cluster.
  2. Edit bluemix/pipeline-DEPLOY.sh at line 60 in your repository and replace helm init with helm init --upgrade.

Then re-run the DEPLOY job.

The toolchain DEPLOY fails with an UPGRADE FAILED error

The DEPLOY log shows:

Error: UPGRADE FAILED: "mmssearch" has no deployed releases

There is a known helm issue. If an install of a given release fails the very first time it was attempted, all subsequent install (upgrade) attempts of that release will fail. To fix, for example in the case of the exact error above related to mmssearch, you can issue a helm delete mmssearch --purge command. This command can be added in the deploy script right before issuing the helm upgrade --install .... command.

Related Content

IBM Cloud solution tutorial: Migrate web applications from Virtual Machines to Kubernetes