Artdaq "wish list" responses » History » Version 6

« Previous - Version 6/11 (diff) - Next » - Current version
John Freeman, 01/12/2017 02:35 PM

Artdaq "wish list" responses

  1. a computer running CentOS7 that has artdaq, lbne-artdaq, nodeJS, MongoDB, artdaq-database, etc. already installed. This computer would be connected to a SLAC RCE with a 10 Gb link, could be connected to the CERN public network (or at least has a well-defined path in and out), and has a moderately large disk that can be used for writing data. [this may be out of our control, but I've included it to set the stage]
  2. an instance of artdaq-database running in MongoDB that has pre-populated entries for
    1. the list of available components: e.g. 1 or more RCEs, 1 or more SSPs, 1 or more FELIX cards
    2. the locations where various BoardReaders should run
    3. the possible locations of EventBuilders
    4. possible location(s) of Aggregators
    5. any other dataflow configuration parameters that are needed
    6. a sample RCE BoardReader and electronics configuration - this can be a shell of what will be needed
      JCF: I can work on getting this into the DB, although one thing I'm not clear on is this: do we need to get the info into the DB using conftool every time we install on a new host, or can we just copy the DB contents over? I should add that right now I have yet to add logic to DAQInterface to make the concept of "possible locations of EventBuilders and Aggregators" meaningful - i.e., DAQInterface doesn't currently care what hosts you specify for the processes, although of course pmt.rb might fail and cause DAQInterface to shut down if you specify hosts to which pmt.rb can't connect.
  3. the ability for an expert user to add another RCE to the list of components or modify the value of a dataflow parameter (mentioned in 2.5)
  4. the ability for an expert user to add parameters to one or more RCE boards in the RCE BR/electronics configuration and the ability to add more boards to the configuration (mentioned in 2.6) [ideally, this would be more sophisticated than "export the FHiCL file, edit it, and import it"]
  5. a user interface that can present the list of available components to a user, when requested (this user interface and the ones mentioned below do not need to be graphical)
    JCF: right now this is as simple as executing "cat .components.txt" from the base directory of artdaq-utilities-daqinterface. Once we put the allowed components list into the database, I imagine we'll need logic to retrieve the list and then cat it to screen.
  6. a user interface that allows a user to pick a set of components to use for a given DAQ session
    JCF: "./bin/ <component1> <component2> ..."
  7. a user interface that allows the startup all of the artdaq processes
    JCF: "./bin/ boot <daqinterface config>"
  8. a user interface that shows the list of available Run configurations (also known as Detector/Electronic configurations) to the user and allows him/her to select one
    JCF: as I understood it from the meeting, this can be done from the command line with conftool
  9. a process that is smart enough to build the FHiCL files that need to be sent to the artdaq processes, given the selected Run configuration and readout components
    JCF: this is handled by DAQInterface using information received during the boot and config transitions
  10. a user interface that allows the configuration all of the artdaq processes
    JCF: "./bin/ config <physics config>"
  11. a (possibly simple) mechanism for automatically assigning unique run numbers
    JCF: The ./bin/ script handles this automatically by looking at the run #'s saved in the run records directory. However, we still have ./bin/ start <run_number>", so I'll remove the <run_number> argument
  12. a user interface that allows a user to begin a run (without having to explicitly provide a run number)
    JCF: See above
  13. archiving of the configuration that is used in each run
    JCF: The FHiCL documents used to initialize the artdaq processes are saved in the *.root files. Even without diskwriting on, data about the run is saved in the run records directory and if we're using the database, saved in the database as well. The script "" in artdaq-utilities-daqinterface can be used to confirm that the FHiCL documents saved in the run records directory and those saved in the output *.root file are the same.
  14. archiving of some amount of run history information, such as begin-run time, end-run time, and number of events in the run
    JCF: still to be implemented
  15. a user interface that allows end run; and later, shutdown
    JCF: ./bin/ stop followed by ./bin/ terminate
  16. as shown on slide 13, the "run control" scripts/application should interact with the Cfg Mgmt system to get the list of available components, etc.
  17. I presume that the Process Manager that is mentioned in the slides will be John's DAQInterface plus pmt.rb
    JCF: correct
  18. The FHiCL Document Creator may need to be created or DAQInterface may fulfill that role
    JCF: handled in DAQInterface using information received during the boot and config transitions