CF Tasks

1. Preface

attendee-service is configured to automatically create a database schema on startup. You can verify this by inspecting the source code of the attendee-service project.

Study the file src/main/resources/application.yml, and note how it contains the setting generate-ddl:

spring:
  application:
    name: attendee-service
  jpa:
    generate-ddl: true

This setting ensures that when the application starts, the schema will be auto generated from the JPA domain definition.

In this lab, we’ll turn off this feature, and instead learn how to run a "one-off" task that will create the needed schema for us explicitly, as a separate step.

2. Setup

If necessary, get a copy of the source code for attendee-service (clone or download)
Edit the application.yml file and set generate-ddl to false. Rebuild the application jar file with:
```
mvn clean package
```
Do whatever’s necessary to delete your existing attendee-service application.
Delete the mysql database you’d originally created for the attendee-service application using the command cf delete-service
Now, install the mysql cf cli plugin from the cloudfoundry plugins web site. This will add the command 'mysql' to your cf cli.
Recreate your attendees database, perhaps like this:
```
cf create-service cleardb spark attendee-db
```
Validate that the service was created by invoking the cf services command.
Push attendee service once more. Assuming you have a manifest file for attendee-service that looks similar to this:
```
---
applications:
- name: attendee-service
  path: target/attendee-service-0.0.1-SNAPSHOT.jar
  memory: 512M
  random-route: true
  services:
  - attendee-db
```
Invoke your cf push command from within the attendee-service base directory:
```
cf push
```
Now, invoke the cf mysql command, which should produce output similar to this:
```
MySQL databases bound to an app:

attendee-db
```
Let’s check that the database has no tables at this time:
```
cf mysql attendee-db
```
then:
```
mysql> show tables;
```
And ensure that the output shows that no tables exist at this time.

In fact, let’s go further:

Tail the logs for attendee-service
```
cf logs attendee-service
```
In a browser, visit the attendees/ endpoint and verify that you get an error, and verify that the logs contain an error message complaining that the attendees table does not exit.

3. Run a "one-off" task to generate the schema

Familiarize yourself with Cloud Foundry tasks by reviewing the documentation.

It turns out that our spring boot application already has the knowledge for creating our schema. That knowledge is embedded in the spring data jpa libraries. The idea is simple:

we’ll run a spring-boot app not as a web service, but as a task by turning off the spring-boot property spring.main.web-environment
next, for this run, we’ll turn on the spring.jpa.generate-ddl property
we’ll specify a start command for our task that leverages everything that’s already inside our droplet: the jdk and our spring boot application

Here’s the command we want to run:

cf run-task attendee-service '.java-buildpack/open_jdk_jre/bin/java \
    -Dspring.jpa.generate-ddl=true \
    -Dspring.profiles.active=cloud \
    -Dspring.main.web-environment=false \
    -cp . org.springframework.boot.loader.JarLauncher'

You’ll see output similar to this:

Creating task for app attendee-service in org eitan-org / space eitan-space as esuez@pivotal.io...
OK

Task has been submitted successfully for execution.
task name:   fb9e9533
task id:     1

Run the 'cf tasks' command to check on the status of our task:

cf tasks attendee-service

The output should resemble this:

id   name       state       start time                      command
1    fb9e9533   SUCCEEDED   Thu, 11 May 2017 02:45:15 UTC   .java-buildpack/open_jdk_jre/bin/java \
    -Dspring.jpa.generate-ddl=true \
    -Dspring.profiles.active=cloud \
    -Dspring.main.web-environment=false \
    -cp . org.springframework.boot.loader.JarLauncher

Note that the state is SUCCEEDED.

Let’s now verify that our table has been created:

cf mysql attendee-db

and:

mysql> show tables;

..should show:

+------------------------------+
| Tables_in_ad_12f26b7197bb693 |
+------------------------------+
| attendee                     |
+------------------------------+
1 row in set (0.03 sec)

Now, attempt to revisit the attendees endpoint once more, and you’ll see that the call succeeds.

Congratulations, you’ve just run a one-off task inside cloudfoundry!

Acknowledgements to Liu Dapeng and his published example.