I took a look at the second case (where we see spots outside the patrol region), and something seems odd with the plots.  When I plot the same data with my scripts, the motion of the cobra starts at the home position, and there are no spots outside the patrol regions; it looks like there's a global offset to the plotted patrol regions. (I updated the pfs_instdata and pfs_utils to the latest master, and checked with Chi-Hung that this convergence did in fact start at home). 

I overplay the most updated cobra centers. I do confirm that most of the cobra started from center. The cobra center in the table cobra geometry is not updated for a while. I will fix this. By adding the updated center, the problem is not the confusion of fiber identification in the overlapping region.

Oh I see, so if Robert used cobra geometry from opdb, that'd explain his strange plots. rhl Can you confirm this ?
I think we always want to be sure that the configuration in opdb match the xml file.
That's a check that could be done automatically by fpsActor when the xml file is loaded, chyan how hard would it be to implement that ?

Closing this ticket, as the ongoing issue of monitoring mis-IDs is subsumed by ticket 1987.