LibraryToggle FramesPrintFeedback

Making Endpoints Stateful

You can configure JMS endpoints to store a copy of the current message exchange in a persistent datastore. This helps in cases where you need to recover from failures.

Overview

Fuse ESB Enterprise JMS endpoints typically do not store any state information. You can, however, configure them to store a copy of the current JMS message being sent. The message can be stored either in memory or in a JDBC configured database.

Having the endpoint store a copy of the current JMS message can aid in recovery from failures. For example, if your application is deployed in a cluster of Fuse ESB Enterprise containers you can configure your endpoints to fail over if one of the containers crashes. If your endpoints are configured to store state in a JDBC database, they can then resend any request that was in process.

Activating statefullness

You configure an endpoint to save a copy of the current message by setting its stateless attribute to false.

Configuring the datastore

By default, JMS endpoints uses a memory based message store. The memory based message store is a simple hash map that is stored in active memory. It cannot persist in the event of a failure, does not support transactions, or access by multiple members of a cluster.

If you need to use a more robust message store, you can configure a provider endpoint to use a JDBC accessible database as a message store. A JDBC message store can be shared among a cluster of endpoints, can be persisted in the event of a failure, and, depending on the database, be enlisted in transactions.

To configure an endpoint to use a JDBC accessible datastore, you configure its storeFactory attribute to reference a bean configuring an instance of the org.apache.servicemix.store.jdbc.JdbcStoreFactory class. Table 8 list the properties you can set for the JDBC store factory.

Table 8. Properties Used to Configure a JDBC Store Factory

NameDescription
clusteredSpecifies if a datastore can be accessed by the members of an endpoint cluster.
transactionalSpecifies if the datastore can be enlisted in transactions.
dataSourceSpecifies the configuration for the data source to be used when creating the store.
adapterSpecifies the configuration for the JDBC adapter used to connect to the data source.

[Note]Note

The values for dataSource and adapter will depend on the database you are using and the JDBC adapter you are using.

Example

The fragment in Example 30 shows the configuration needed for a stateful JMS provider endpoint using MySQL as a JDBC accessible datastore.

Example 30. Configuring a Statefull JMS Provider Endpoint

<jms:provider service="tns:widgetServer"
              endpoint="widgetPort"
              storeFactory="#storeFactory"> 1
              stateless="false" /> 2

<bean id="storeFactory" 3
      class="org.apache.servicemix.store.jdbc.JdbcStoreFactory">
  <property name="clustered" value="true"/>
  <property name="dataSource">
    <ref local="mysql-ds"/>
  </property>
</bean>

<bean id="mysql-ds" 4
      class="com.mchange.v2.c3p0.ComboPooledDataSource"
      destroy-method="close">
  <property name="driverClass" value="com.mysql.jdbc.Driver"/>
  <property name="jdbcUrl"
            value="jdbc:mysql://localhost:3306/activemq?relaxAutoCommit=true"/>
  <property name="user" value="activemq"/>
  <property name="password" value="activemq"/>
  <property name="minPoolSize" value="5"/>
  <property name="maxPoolSize" value="10"/>
  <property name="acquireIncrement" value="3"/>
  <property name="autoCommitOnClose" value="false"/>
</bean>

The fragment in Example 30 does the following:

1

Configures the endpoint's store factory by providing a reference to the bean configuring the factory.

2

Configures the endpoint to store a copy of the current message in the datastore.

3

Configures the JDBC factory store to create a datastore that can be accessed by a cluster of endpoints.

4

Configures the MySQL JDBC driver.

Comments powered by Disqus