Skip to content

Latest commit

 

History

History
180 lines (134 loc) · 6 KB

README.md

File metadata and controls

180 lines (134 loc) · 6 KB
Raw

Deploy an application with IBM Watson services

In this section, you will deploy an application that leverages the Watson Speech To Text service.

Create the Watson Speech To Text Service (Skip if the Watson API key is provided for you)

In this section, you will create the Watson service in your own account and get the service credentials.

  1. Go to cloud.ibm.com and using the account switcher drop down, switch to your own Account.
  2. Click on the Catalog, search for Speech To Text service and Create
  3. Click on Service Credentials on the left
  4. Click the dropdown arrow on the credentials to expand it.
  5. Take note of the apikey value. You will need this key in the next section

Update the credentials.json file

  1. Change to the Exercise 4 directory:

    cd ../exercise-4

  2. Edit the credentials.json file found in watson/credentials.json.

nano watson/credentials.json

  1. Update the apikey value with the value from the previous section and then save the file.

  2. Create a Kubernetes Secret from on the credentials stored in this file.

    kubectl create secret generic apikey --from-file=./watson/credentials.json

  3. List secrets to see your secret called apikey.

kubectl get secret

Build the image for the Watson Speech to Text application

  1. Create a unique application name for the watson app you're about to build. Something like bmv-watson-1111

    export MYWATSONAPP=<your_unique_name>

  2. Build the image, and push it to the IBM Cloud Container Registry using the ibmcloud cr build command.

    ibmcloud cr build -t $MYREGISTRY/$MYNAMESPACE/$MYWATSONAPP ./watson

  3. List images to find your new image. You will need the image name in the next step.

ibmcloud cr images
  1. Edit the watson-deployment.yml file and update the image tag with the registry path to the image you just created.
nano watson-deployment.yml

Here is the relevant part for you to edit: yml spec: containers: - name: watson-speech-to-text image: "us.icr.io/<namespace>/<appname>" # edit here! # change to the path of the watson image you just built and pushed # ex: image: "us.icr.io/code-camp/bmv-watson-1234"

Share the Watson API Credentials with your application

Take a close look at the watson-deployment.yaml and look for the volumeMounts and volumes section. This is how we tell Kubernetes to share the apikey secret we created with the container. It writes the contents of the secret into /var/credentials of the container.

volumeMounts:
  - mountPath: /var/credentials
    name: apikeyvol
...
volumes:
  - name: apikeyvol
    secret:
      secretName: apikey

There is nothing you need to do for this section.

Create deployments, pods and services

The yaml file has everything we need defined. We simply give this file to kubectl to apply the configuration.

  1. Create the deployment, pods and services making up your application using the .yml file.

    kubectl apply -f watson-deployment.yml

  2. You can open the Kubernetes dashboard and explore all the new objects created or use the following commands.

    kubectl get pods,deployments,services

Expose this application to the public internet using the IKS Ingress ALB

In the previous sections, you exposed applications using service type NodePort. NodePort is good for development purposes, but when you are ready to go live, you need a permanent IP address or host name. For that, you need to create a service type of either LoadBalancer or create a Kubernetes Ingress.

Standard clusters on IKS come with an IBM-provided domain. This gives you a better option to expose applications with a proper URL and on standard HTTP/S ports. In this section, we'll use Ingress to set up the cluster inbound connection to the service.

  1. Get the Ingress information for your cluster

    ibmcloud ks cluster get --cluster $MYCLUSTER

    Example output:

    Name:                           mydemocluster
ID:                             f4f207d35a2b4fe98998d7ba0d
State:                          normal
Created:                        2018-12-06T16:11:59+0000
Location:                       dal13
...
Ingress Subdomain:              mydemocluster.us-south.containers.appdomain.cloud
Ingress Secret:                 mydemocluster
...

  2. Note the Ingress Subdomain and Ingress Secret values. You'll need this in the next step.

  3. In watson-ingress.yaml, update the three locations marked <Ingress Subdomain> and <Ingress Secret>.

nano watson-ingress.yaml

Here is the relevant portion you need to edit:

spec:
  tls:
  - hosts:
    - watson.<Ingress Subdomain>
    secretName: <Ingress Secret>
  rules:
  - host: watson.<Ingress Subdomain>
  ```

  ![](../README_images/watson-ingress.png)

4. Apply this yaml to your cluster.
  ```
  kubectl apply -f watson-ingress.yaml
  ```
4. In a new browser tab, go to your application! The URL will be `https://watson.<Ingress Subdomain>`


![](../README_images/watson-stt.png)

You should be able to click the **Record** button and start speaking into your microphone. Watch the text get trascribed live!

## Clean up

1. Clean up the deployment, pods, and services you created:

  ```
  kubectl delete -f watson-deployment.yml
  ```
2. Clean up the Ingress you created:

kubectl delete -f watson-ingress.yaml


3. Clean up the images in the registry.

ibmcloud cr image-rm $MYREGISTRY/$MYNAMESPACE/$MYWATSONAPP

4. If you created a Watson Speech to Text service at the beginning of this exercise, return to the browser tab where it is open. Click `Actions > Delete Service` and then click 'Ok' to delete the service.

Continue on to [Exercise 5](../exercise-5/README.md)