This site contains the documentation that is relevant to older WSO2 product versions and offerings.
For the latest WSO2 documentation, visit https://wso2.com/documentation/.

Setting Up Delayed Delivery

This explains how you can set up delayed delivery in the broker.

Setting up delayed delivery for topics

When a topic has one or more subscriptions and when you need to apply a global delay to all the subscriptions of it, you can set it up in the broker by defining global policies.

Setting up delayed delivery for subscribers

The client who connects can override the global policies you set up for topics. When configuring the java.naming.provider.url parameter in the JMS Inbound Endpoint, you can pass the re-delivery policy property as a query String. For a detailed description on the policy properties, go to Apache ActiveMQ Documentation.

For example, when configuring a Script mediator, if you need to rollback the message four times upon an error with a one second delay, set the java.naming.provider.url property as follows.

Script Mediator configuration
tcp://localhost:61616?jms.redeliveryPolicy.redeliveryDelay=1000&jms.redeliveryPolicy.maximumRedeliveries=4

The redeliveryDelay property specifies the re-delivery delay in milliseconds and the maximumRedeliveries property defines the number of times the retry should occur. For more information on these properties, go to Apache ActiveMQ Documentation.

In ActiveMQ 5.9.0, although the broker properly identifies the re-delivery times, the broker does not delay the messages as expected. Thus, it will not enforce even the global level delay.

The final Inbound Endpoint configuration with the re-delivery delay and number of times you defined will be as follows:

Inbound Endpoint configuration
<inboundEndpoint name="MARSInboundEP" onError="MARSEPErrorSeq"
        protocol="jms" sequence="ProcessOrderSeq" suspend="false">
        <parameters>
                .......
               <parameter name="java.naming.provider.url">tcp://localhost:61616?jms.redeliveryPolicy.redeliveryDelay=1000&amp;jms.redeliveryPolicy.maximumRedeliveries=4</parameter>
                 ......
        </parameters>
</inboundEndpoint>

Enforcing a delay per message

There are not an out of the box feature available in ActiveMQ to enforce delay per message. The re-delivery policies are defined by the Initial Context Factory when it establishes the connection and it does not happen on a per message basis.

However, before rolling back, you can enforce a delay using the Script mediator in the error sequence  as follows.

Script Mediator configuration
<script language="js">java.lang.Thread.sleep(10000);</script> 

Then, the sample sequence is as follows.

Sample Sequence configuration
<?xml version="1.0" encoding="UTF-8"?>
<sequence name="MARSEPErrorSeq" xmlns="http://ws.apache.org/ns/synapse">
    <log level="full">
        <property expression="get-property('TxBID')" name="TxBID" xmlns:ns="http://org.apache.synapse/xsd"/>
        <property expression="get-property('MARSGUID')" name="MARSGUID" xmlns:ns="http://org.apache.synapse/xsd"/>
        <property expression="get-property('AppName')" name="AppName" xmlns:ns="http://org.apache.synapse/xsd"/>
        <property name="Method" value="MARSEPErrorSeq"/>
    </log>
    <property name="SET_ROLLBACK_ONLY" scope="default" type="STRING" value="true"/>
    <script language="js"><![CDATA[java.lang.Thread.sleep(10000);]]></script>
</sequence>

The Script mediator holds the end of the sequence for the defined number of times before it reaches the end. The rollback() operation occurs once the sequence ends. Hence, this holds onto the rollback operation for a specified period of time. Further, this re-delivers the message only after the specified duration.

Therefore, in the above example, the java.lang.Thread.sleep(10000) property schedules the message back after 10 seconds.

There is no grantee that the re-scheduling will occur as soon as the defined time period in the sleep(..) operation ends. Also, it considers the delay of any re-deliveries. For example, if the re-delivery delay is x seconds and the sleep time is y seconds, then the overall time for the message to get re-delivered is x+y seconds.

Also you can use the filter mediator and switch mediator to do the filtering as shown in the example below, when you need to define re-delivery intervals based on the message error condition dynamically.

Switch Mediator configuration
<switch source="">
   <case regex="string">
      <script language="js"><![CDATA[java.lang.Thread.sleep(10000);]]></script>
   </case>
   <default>
      <script language="js"><![CDATA[java.lang.Thread.sleep(2000);]]></script>
   </default>
</switch>