Nest Labs Fault Injection (nlfaultinjection) is designed to provide a simple, portable fault injection framework.
It should be capable of running on just about any system, no matter how constrained and depends only on the C Standard Library.
This is a simple framework that helps implement fault injection APIs. Fault injection APIs simplify the testing (manual or automated) of the error handling logic of individual components and whole systems; as a testing tool, a fault injection API is not as powerful as an extensive set of mock objects, but has the advantage of being able to run on a live system.
This is not an official Google product.
A SW module implements a fault-injection API by doing the following:
- Instantiate a FaultInjection::Manager object and define a list fault IDs; the Manager needs a name, storage for the configuration of the faults and the name of each fault; the Manager instance is the public API, through which another module of the system can configure faults (for example a test harness can turn on a fault by calling the FailAtFault() method). The module should also provide a public function that returns a C++ reference to the Manager and initializes the Manager the first time it is called.
- At the location in which the fault is to be injected, the nlFAULT_INJECT macro is used to inline code that if executed triggers a fault in the system; the macro queries the Manager object calling the CheckFault() method and executes the extra code if it returns true. The invocations of CheckFault are counted per faultID, which allows a test harness to automatically loop over different valid configurations.
There are numerous avenues for nlfaultinjection support:
- Bugs and feature requests — submit to the Issue Tracker
- Google Groups — discussion and announcements
- nlfaultinjection-announce — release notes and new updates on nlfaultinjection
- nlfaultinjection-users — discuss use of and enhancements to nlfaultinjection