AndrewEdmonds11
commented
2 months ago So I'm still seeing this issue and having dug in a bit deeper I think it is somewhere in or underneath the DQMHistoConsumer. The bug is consistent with a buffer not being cleared somewhere. I am sending one event per second into artdaq/otsdaq and also one histogram per second in HistoSender.
HistoReceiver is (obviously) being called more frequently but always reads something from the buffer. The histogram I see in the visualizer is just the first histogram I send being added to itself much more frequently.
It looks like the only time that the code enters the if statement below is before I send events into artdaq/otsdaq. Afterwards, it's always getting to the readPacket part.
One last piece of information that may be helpful: I'm running everything on a single node, and Giani has reported he does not see the issue when running on different nodes.
Documenting this bug I saw a while back. The otsdaq visualizer is not receiving the histograms that are made in the STMDQM module. In the picture below, the histogram on the right is what is made in the STM DQM module. It is just plotting the trigger numbers so should be a uniform distribution that goes up to 1 on the y-axis.
The histogram on the left is what appears in the visualizer. This histogram is the one that auto-updates during running and just keeps on adding entries to the first bin. Note that the y-axis goes up to 900e3.