How to handle a hung flow
DavidBrickell 270001MT2R Visits (5067)
Sometimes when developing a flow we find that the flow becomes non-responsive. This can be caused by a few different reasons but one of the most common is that the flow is stuck in a logic loop.
This can be caused by poor code design or design of the flow itself. Even after identifying and fixing the logic within the flow, you are still stuck with the task of replacing the deployed flow within the Broker.
WebSphere Message Broker has built in recovery measures. This means, if the Broker detects that the EG has crashed or stopped unexpectedly, it will try to restart this process. To avoid the flow going back into its hung state, you should shut down the input source. For example, you can disable the input queue if using MQ as your input source. After this is done, you can kill the EG process. When the EG restarts you will see an error stating the message flow couldn't connect to the input source. From this point you can stop the message flow and then delete it.
This will allow you to redeploy the corrected flow to replace the old one.