Hi,
I have a question on an ESB proxy/sequence. I will do my best describing the
problem and the issues I'm facing:
Let's assume that for this case we have 1 ESB and 2 AS. The AS are copies of
each other. They provide load-balancing/ High availability. On the ESB we have
one proxy which will call endpoints on the AS.
ESB proxy
|
|-sequence
|
|-- EP1 --( o-- AS1
|
|-- EP2 --( o-- AS2
It is important that following constraints are met at all times:
- 1 invocation of a soap method on the ESB will result in 1 invocation on
each AS (given they're up and running)
- When both servers return a valid response (Not soap fault), one of the
two responses should be the ESB response
- When one of the two AS returns an error/fault, the response from the
ESB proxy should be this error/fault
- When one of the two AS is down, we should invoke the endpoint that is
still alive and the response (valid or fault) should be the response of the ESB
proxy.
The setup highly resembles the Scatter-Gather EIP; we have one request, we
scatter it over our AS endpoints, we gather the responses and based on some
rules we determine the ESB response. We just lack a solution to handle the
error that occurs when an endpoint is down.
I already configured multiple ESB sequences to try to make this happen as it
should. The problem is that when an endpoint is down, we get a "could not
connect to back end" error, as expected. This error can only be handled in the
fault sequence. We cannot receive the soap fault into our gather step to cope
with it.
Is there a way to achieve this?
How do you advise on configuring this?
Kind Regards,
Koen Geerts
_______________________________________________
Dev mailing list
[email protected]
http://wso2.org/cgi-bin/mailman/listinfo/dev
