Checking asynchronous applications
The examples here have considered simple synchronous, single-threaded applications. In multi-threaded and asynchronous applications, different code sections can run out of sequence or in parallel depending on external events. While you can test that each function works correctly, it is much harder to test their interactions. You can’t define a single-state machine, and it’s best to find bugs with load testing. That is described in detail in Chapter 12, Load Testing.
Asynchronous applications involve sending messages from a subsystem of an application and pausing further processing or continuing with other tasks until it receives responses. That encompasses a vast range of possible uses, but there are a common set of checks to apply to them, as follows:
- Duplicate messages
- Out-of-order messages
- Missing update messages
- Missing response messages (timeouts)
This section considers each of these cases.
