CloudFoundry Service Connectors

This set of (Maven) projects provides bare-bones samples of Cloud Foundry server connectors. That is: Spring framework dependent functionality that allows for a convenient approach to Spring datasource construction (auto reconfiguration) as performed by the standard Cloud Foundry Java buildpack. For a detailed description of this functionality, please consult https://github.com/spring-projects/spring-cloud and https://github.com/spring-projects/spring-cloud/tree/master/spring-service-connector.

Usage

To make use of these connectors, do the following:

• Check out these Spring service connector projects (we'll assume for the remainder of this discussion that we want to use the Hawq (*) connector)
• Build the connector project and install to your local Maven repo
• Add a dependency to the Hawq service connector and associated libs to the application project
• Make sure to define a 'bare bones' Spring datasource in the application's 'cloud' profile in its Spring application context
• Define a "user provided" service instance for the Hawq deployment to be targeted
• Push the application and bind the service instance to the application

(*) Note that the Hawq connector allows for connecting our application to an (external) Hawq deployment. It uses / depends on a PostgreSQL JDBC driver.

####Build and Install Hawq Spring Service Connector

$ cd <actual-hawq-spring-connector-directory>
$ mvn clean install

This will result in the Hawq Spring connector deployable (jar) ending up in ~/.m2/repository/org/cloudfoundry/hawq-cloud-db-connector/1.0.0-BUILD-SNAPSHOT ####Add dependency to Hawq Service Connector and Associated Libs Assuming our application code is structured as a Maven project, add the following dependencies to the applications project's pom.xml:

<!-- HAWQ service connector -->
<dependency>
    <groupId>org.cloudfoundry</groupId>
    <artifactId>hawq-cloud-db-connector</artifactId>
    <version>1.0.0-BUILD-SNAPSHOT</version>
</dependency>

As we need a PostgreSQL driver, also add:

<dependency>
	<groupId>org.postgresql</groupId>
	<artifactId>postgresql</artifactId>
	<version>9.2-1003-jdbc4</version>
</dependency>

As we rely on the standard Pivotal Cloud Foundry Java buildpack's role in 'auto reconfiguration', make sure the following dependencies are also present:

<!-- Spring Cloud -->
<dependency>
    <groupId>org.springframework.cloud</groupId>
    <artifactId>spring-service-connector</artifactId>
    <version>0.9.2</version>
</dependency>
<dependency>
    <groupId>org.springframework.cloud</groupId>
    <artifactId>cloudfoundry-connector</artifactId>
    <version>0.9.2</version>
</dependency>

####Definition of DataSource in Application's 'cloud' Profile The datasource that "represents" the application's connection to Hawq can be defined as follows in the application's Spring application context (assuming XML):

<beans profile="cloud">
    <cloud:data-source id="dataSource" service-name="hawq">
        <cloud:connection properties="sessionVariables=sql_mode='ANSI';characterEncoding=UTF-8"/>
        <cloud:pool pool-size="20" max-wait-time="200"/>
    </cloud:data-source>
</beans>

Note that we are assuming that the name of the 'user provided' service instance - to be defined below - is 'hawq' Also note that the listed cloud:connection properties are not compulsory, but provide an example of what can be specified ####Define a "user provided" service instance for the Hawq Deployment A key aspect to take into account is the way in which the auto reconfiguration functionality triggers the use of our Hawq Spring service connector. Inspecting the code (specifically: HawqServiceInfoCreator.java) will show that this is based on the (partial) match of a service instance property with the string "hawq" (note that one of those properties is the name of the service instance itself...). The interaction with the Pivotal Cloud Foundry CLI to to define the service instance consists of:

$ cf cups hawq -p "host, port, user, password, name"

host> <actual-hawq-deployment-hostname-or-ip-address>

port> <actual-hawq-port>

user> <actual-hawq-user-name>

password> <actual-hawq-user-password>

name> actual-hawq-db-name
Creating user provided service hawq in org actual-org-name / space development as [email protected]...
OK
$

####Push and Bind the Application Finally, we can push and bind the Hawq service instance.

$ cd <actual-application-project-dir>
$ cf push <application-name> -p <location-of-application-deployable> --no-start
$ cf bind-service <application-name> hawq
$ cf push <application-name> -p <location-of-application-deployable>

At this point the application should have a a valid Hawq-connected datasource to make use of...