Hi All, At the moment, current MFP implementation does not support retrying based on HTTP status codes. It only supports retrying for transport level failures but not for application level failures. As you know, in HTTP protocol, application level failures are distinguished using different status codes. In the case of MFP, it is important to be able to retry for 5xx server errors. But there also could be rare occasions in which retrying for 4xx and 3xx are useful. Especially when dealing with servers that do not follow HTTP protocol exactly as it is.
Therefore, I have improved MFP implementation to support retrying based on the returned HTTP status code as well. Say, a user wants to retry for HTTP status codes 500 and 504 but not for any other HTTP status code. In that case user can configure the MFP as below. Please note the parameter in bold font which I have introduced along with this feature. <messageProcessor xmlns="http://ws.apache.org/ns/synapse"; class="org.apache.synapse.message.processors.forward.ScheduledMessageForwardingProcessor" name="Processor2" messageStore="JMSMS"> <parameter name="max.delivery.attempts">4</parameter> <parameter name="message.processor.reply.sequence">replySequence</parameter> <parameter name="interval">1000</parameter> *<parameter name="retry.http.status.codes">500, 504</parameter>* </messageProcessor> Please not that the above parameter does not come to effect in case of out-only. Implementation of this feature can be found in [1]. Once this is merged, I am planning to send another PR for the required documentation changes. [1] https://github.com/apache/synapse/pull/32 Thanks, Shafreen
