Trigger Replay Application¶

The "replay_tps" python application is a tool to 'replay' TPs through a configurable trigger application, emulating a 'normal' trigger flow.
This is achieved by having a custom version of fddaqconf, replacing readout with TriggerPrimitiveMaker plugin(s). It accepts TPstream HDF5 file(s). This app only creates the configuration. NanoRC is then used to run the system.

Configuration¶

It's run using python -m replay_tps. Additional arguments: - -s / --slowdown-factor: [defalt=1] Can slow the flow of time by dividing the clock frequency, used by by TriggerPrimitiveMaker [TPM] - -l / --number-of-loops: [default='-1'] Number of times to loop over the input files (-1 for infinite) by [TPM] - -to / --tpset-time-offset: [default=0] A time offset applied to instances of [TPM] - -tw / --tpset-time-width: [default=10000] A width of the window used to pack TPs to TPSets by the [TPM] - -wt / --maximum-wait-time-us: [default=1000] A litte time buffer to wait between sending TPSets, used by the [TPM] - -f / --input-file: [*required] TPStream HDF5 file to be replayed - -map / --map-file: [] DRO map file - -n / --number-of-links: [default=1] If DRO map not provided, replay will generate one for this number of links - --debug: [flag] Switch to get a lot of printout and dot files - json_dir: The name of the output directory that will hold the configuration files

Example commands:¶

python -m trigger.replay_tps -f <tpstream_file.hdf5> -map <dro_map.json> <name_of_output_config_folder>

Alternatively, one can ommit providing a DRO map file. In this case a map will be generated internally for the n number of links specified. For example to use 4 links:

python -m trigger.replay_tps -f <tpstream_file.hdf5> -n 4 <name_of_output_config_folder>

Finally, one can choose to also provide a custom trigger configuration (specifying makers), otherwise the defaults are use:

python -m trigger.replay_tps -f <tpstream_file.hdf5> -c <config.json> <name_of_output_config_folder>

Notes:¶

the same data (same file) is passed through each link (can be offset)
the tpsets are 'custom' made by the TPM
no readout = no readout buffers
several applications are 'missing': timing, hsi, tpsetwritter...
(automated) integration test using replay for new PRs

Details:¶

This application consists of 2 scripts: - __main__.py: the custom version of fddaqconf_gen - replay_tp_app.py: the definition of the replay application

main.py¶

This script generates dunedaq configuration. Also calls the replay_app from replay_tp_app.py which acts as a standalone dunedaq application. Otherwise mostly follows fddaqconf_gen but is a stripped-down version. Functionality: - check directories - prepare debug if configured - load the provided configuration, if none provided load the default one - expand the configuration: currently only using boot, detector, daq_common, readout, trigger, dataflow - validate configuration - check input HDF5 file - prepare the DRO map, either check the one provided or generate one given the number of links configured - create the appropriate applications: dataflow, replay, trigger, dfo (notice that there is no readout app at this point, while it is used to prepare theSystem object, it is then replaced by the replay application) - prepare boot, connect fragment producers, prepare MLT links - prepare per-app command data - write final configuration and metadata

replay_tp_app.py¶

This script defines the dunedaq replay application, replay_app. The configuration is passed from __main__.py. Parameters: slowdown_factor, number_of_loops, n_streams, time_offset, time_width, wait_time. Functionality: - sets up variables (such as the clock frequency) - creates tp streams: the number depends on the links of DRO map; these 'pass' the HDF5 tpstream file data - creates TriggerPrimitiveMaker module - modifies the mgraph (setting the connections between the TPM and tp streams)