Closed mroda88 closed 2 years ago
To test this branch, the corresponding PRs need to be obtained from appfwk, dfmessages, dfmodules and trigger
Hi, all the tests that I have run are ok (except the one with DQM, but we expect these to fail), the code looks correct. So as far as I'm concerned this looks good.
For reference. With 3 RU apps, 3 DF apps, trigger at 1 KHz, number of tokens = 10, free watermark = tokes/2 = 5, the system saturates around 450 Hz.
At the same time with the changes in the DFO to MLT busy signal, the rate of the busy signal transmission is around 40 - 50 Hz.
I think this addresses both the points made today. I'd like to have this run go on for a few hours before merging though.
I have a nigh-long run. here are some relevant plots, of the last 12h of the run. As you can see the DFO activity is stable. The accepted trigger rate is around 400 Hz, while the generated triggers were at 1 KHz. All the branches across the repos are also synced with develop, so I think we can merge whenever we want.
The test was done on the new CentOS machines.
This PR changes the configurations to allow the DFO to live in its own application.