This is a simple prototype to help prove the concepts behind the proposed new "Convergence" architecture for OpenStack Heat.
It uses a trivial in-memory datastore and an event loop within a single process to simulate multiple processes with asynchronous message passing. Note that it cannot simulate race conditions or other concurrency errors, since only one event is ever processed at a time.
Run the code by typing python -m converge. Note that when run directly like
this, the test assertions in the scenarios are ignored (a warning is logged).
This mode is useful for debugging and becoming familiar with the code because
it displays all of the log messages that allow you to follow the execution.
Usage: /usr/bin/python -m converge [options] SCENARIOS
An OpenStack Heat convergence algorithm simulator.
Options:
-h, --help show this help message and exit
-d DIR, --scenario-directory=DIR
Directory to read scenarios from
-p, --pdb Enable debugging with pdb
To run a specific subset of the test scenarios, append their names to the
command line (e.g. python -m converge basic_create).
To automatically set debugger breakpoints at the beginning of each asynchronous
task, pass the --pdb option on the command line.
The unit tests can also be used to not only run the scenarios but verify the correctness of the results. To do this, run:
python -m unittest test_converge
You can also use some other testrunners, including and maybe not even limited to Nose:
nosetests test_converge