Under certain conditions
message flow processing
can become unresponsive, for instance when:
- Waiting for a response from an external system.
- Processing an infinite loop or a calculation that takes a very
long time.
- Deadlocked between two resources.
In previous releases of
WebSphere® Message Broker the
only way to resolve this situation is by terminating the integration
server process. As part of
IBM® Integration Bus Version 9.0 new
functions are introduced to allow termination of unresponsive
message flows in a controlled manner.
The following topics explain the functions available:
- Programmatically check from within a message flow if it has been requested to
stop
By using the programming APIs (ESQL, Java™, and .NET), you are able to see if a message flow has been requested to stop.
For more information, see Check if a message flow has been requested to stop.
- Manually force a message flow to
stop
A force option –f has been added to the mqsistopmsgflow command
to allow you to specify how the message flow can
be forced to stop. For more information, see Manually force a message flow to stop.
- Automatically force a message flow to
stop
Two message flow properties are provided which can be set within
the BAR file and the
workload management policy:
- processingTimeoutSec – maximum
time a message flow can process a message
before taking an action.
- processingTimeoutAction – the action
to take. Currently, this is restricted to none or restartExecutionGroup.
Additionally, an event message is published on a WebSphere MQ topic once the
processingTimeoutSec is exceeded and again when
the
message flow processing has finished.
For
more information, see Automatically force a message flow to stop.
Last updated Friday, 21 July 2017