MQSeries.net :: View topic - Question about failure handling

jharringa · Posted: Wed Nov 28, 2007 4:00 pm Post subject:

I have the following non-transactional flow:

MQIn -> Compute1 -> SubFlow1 -> SubFlow2 -> Compute2 -> MQReply

SubFlow1 looks like this:

Input -> Compute3 -> MQOut (Request) -> Compute4 (Move MsgId to CorrelId) -> MQGet on CorrelId -> Compute5 -> Output

SubFlow1 has 2 Outputs:
1 for normal processing
1 for Failure processing in the subflow

The failure terminals of Compute3, MQOut, MQGet, and Compute5 are wired to the Failure Output in the Subflow and both Outputs for SubFlow1 are wired to the input of SubFlow2. The Warning and No Message terminals of the MQGet on CorrelId are each wired to their own Throw node.

SubFlow2 looks similar to SubFlow1 but it only has 1 Output and all Failure terminals are unwired.

The intent here is that if the processing in SubFlow1 fails then I want to continue on to SubFlow2.

My problem is thus: The handling is working correctly in SubFlow1 and it will continue no matter what. However, if there is a problem in SubFlow2 then the failure seems to get caught by the failure terminal of my MQGet on CorrelId within SubFlow1. Since that is wired back into SubFlow2 my flow continues again into SubFlow2. If I encounter the error in SubFlow2 again then the flow is brought back to my original MQInput as I had expected before.

So, with that, am I seeing odd behavior or is this to be expected? Is there a better way of gracefully handling failures in SubFlow1?

I am running Broker 6.0.0.5 and Toolkit 6.0.2 ifix 009

By the way, I know that this is orchestration that shouldn't be done in the Broker. However, we are migrating from ICS to Broker. We don't have WPS to do the orchestration and we are getting away from using ICS.