During startup, Heat incorrectly ignored nested stacks when searching for stacks with interrupted operations (for example, ones form a previous heat-engine process exiting). In addition, while those stacks that were not ignored were correctly set to FAILED, their resources remained IN_PROGRESS.
Because the resources remained IN_PROGRESS, it was not possible to recover the stacks when heat-engine was restarted.
With this update, nested stacks are now included when searching for interrupted operations, and IN_PROGRESS resources as well as stacks are moved to the FAILED state. Consequently, they can be recovered as expected upon restart of heat-engine.