Notes from protoDUNE SP DAQ review November 2016

04-Nov-2016, KAB

Short-term to-do items:
  • send instructions for the TriggeredFragmentGenerator demo to Frank Filthaut [done, 11-Nov]
Things to think about:
  • Both from discussions with Frank and Wes, it seems like an artdaq developer's workshop would be helpful. To make this most useful, we will need to develop a step-by-step tutorial that demonstrates some of the most likely needs of developers. And, the artdaq-skeleton would be useful for this.
  • Reliable sending (TCP) of trigger/pull messages is expected.
  • Decoupled event flow and metrics gathering/reporting is expected.
  • How do we easily support different system configurations, for example, data logging from AGs vs. EBs? A graphical tool for system (i.e. architecture configuration) design?
  • When/if do we develop the routing master (aka dataflow orchestrator)?
  • What ideas do we have for transferring dataflow metrics to the TriggerBoard FragmentGenerator for trigger throttling?
[KAB] What do we need to do in the short/medium term?
  • review the talks from the review and understand the timeline
  • Redmine Issues and release plan
  • talk to people about the merging of the BI and CRT data
  • learn and/or influence the design of the DAQ cluster (servers, networking, disks)
  • it would probably be good to split up our configuration step before trying to integrate with JCOP
To-do list:
  • test on CentOS7
  • provide a tarball that Giovanna can use for testing artdaq disk writing on an Atlas disk appliance (available after 07-Dec)
  • define desirable feature sets and create a development/release plan
Suggestions from Wes on the tutorial:
  • art and artdaq FHiCL parameters explanation; for example, are "endPaths" optional or required?
  • how does one make scripts to drive the system and monitor its state?
  • we could create a "toy" experiment, including a test driver
  • how does one update artdaq versions and where does one find the list of ones available?