Last night I monitored my show for 2 of the 4 hours it is running, and the problem occurred 3 times. The lights went completely dark during random song, at random points in the songs, while the music will keep playing. I had left the control panel attached, and when the problem happened the second time, I went to 'controls' and paused the dark show. The music paused correctly, and the lights were already OFF. When resuming, the music resumed, but the lights stayed dark.
In each cases, when the next song started, the lights came back correctly.
Without more insights into the code, it is difficult to pinpoint what and from where the problem is coming from, but based on the descriptions of how it occurs, it seems that in some instances, the streaming from the CF card of the .seq file stops, and the system keeps sending packets full of zeroes to the HUBs.
It would be helpful if we could get a feedback of the streaming of the .seq file in the control panel, similar to the time information we are getting for the current performance (something like .seq file read pointer at:xxxx out of yyyy, read errors:zzzz. Last read checksum:aaaa). Having a misbehaving system myself, I would/could do extensive testings on this. That would not tell us why it stopped streaming, if it does, but bring us closer to an understanding, and a solution of this problem.
For tonight show, I will do the same manipulation, without having the control panel opened (well without having the etherdongle connected to the network). Just in case random noise on the network ending up 'confusing' the etd/conductor.
If that doesn't work, tomorrow I will try with a reformatted CF card (the formatting is correct on this one, but some files have been erased/replaced) and a brand new set of files.
If that doesn't work, I will use a different scheduler for the rest of the season.