Running

The live data prototype consists of several python scripts that run independently (and communicate via ZeroMQ). Certain aspects (typically parameters) of these components can be controlled via a ZeroMQ interface.

There are 4 main components that need to be started:

• Streamer (generates fake event data)
• Backend-live-listener (reads event stream, distributes it to all MPI ranks, and buffers event data)
• Backend-reducer (reads data from live-listener and runs Mantid-based data reduction)
• Visualizer (lightweight visualization of reduciton result)

The respective commands are:

Streamer:

python run_streamer.py

Backend-live-listener:

mpirun -n <N> python run_backend_event_listener.py

where <N> is the desired number of MPI ranks. mpirun -n <N> can be omitted to run only with a single process.

Backend-reducer:

mpirun -n <N> python run_backend_reducer.py

where <N> is the desired number of MPI ranks. It must match the number of ranks used for the backend-live-listener.

Visualizer:

python run_visualizer.py

In all four cases the run script supports the option -h to print help, and further options, as described in the respective help. In particular, in several cases you can set the log level with -l and the host with -H (which defaults to localhost).

Control while running:

...is possible with the script control.py. The components described above listen for commands via ZeroMQ on various ports (streamer on 10002, backend-reducer on 10004 and 10005).

Usage example:

# Query server for available controllees and commands
python control.py -p <PORT>
# Example usage for setting the bin size for the backend-reducer
python control.py -p 10005 -c BackendMantidReducer bin_parameters '0.4,0.001,5'

Running the unit tests

python -m unittest discover tests

or for individual modules

python -m unittest discover tests.test_backend