artdaq communication breakdown in an event pull mode ?

[pavel1murat]

I'm trying to run artdaq in an event pull mode [(TRK+CAL) + CRV] using artdaq-demo -based boardreaders. Artdaq is configured as follows:

1. subsystem 1: two boardreaders, BR11 ad BR12, and one event builder , EB06
2. subsystem 2: one boardreader, BR21 ,running as a receiver of multicast requests, one event builder, EB11, running as a sender of multicast requests, and a data logger, DL01.

when EB11 receives an event from EB06, it is supposed to send out a multicast request. BR21 is supposed to receive that request and send a fragment of the requested event back to EB11.

all processes , except DL01, are running on mu2e-trk-06, DL01 is running on mu2e-dl-01.

The tested configuration is running at a rate of 5 Hz of MC-generated input.

After about 30 events, BR21 stops sending fragmenrs to EB11, and only 30+/1 first events in the output file contain fragments sent by BR21.

The symptoms are consistent with the breakdown of the artdaq IPC.

the ardaq FCL files of the processes involved can be found in mu2e-dl-01:/scratch/mu2e/mu2etrk/daquser_002_v001/run_records/120352

Is there a validated example of artdaq demo configuration running in a pul mode ?

-- regards, Pasha