Aaron Roodman had noticed truncated donuts on some images, suggesting the dome or wind screen blocking part of the telescope beam.
Marisa March organized an eyeball squad to look at donut images for SV exposures provided by Kevin Reil, to mark those which show signs of occlusion. She reports here and here. There are clumps of occluded images at larger zenith angles and a small group near zenith.
Gary Bernstein has taken the exposures reported as heavily occluded (>30% on either intrafocal or extrafocal donuts) and calculated the orientation of the occlusion to be expected if it results from the wind screen vs one of the edges. Here are the results. The color of a dot encodes the fraction of occlusion reported by the inspectors. The red/green/blue bands are the orientations of occlusion expected from the (left edge of dome slit) / (wind screen at bottom) / (right edge of dome slit).
The extrafocal donuts seem to be oriented where the edges of the slit would produce problems. There is no sign of wind screen occultation.
The intrafocal donuts are all reported with nearly the same orientation of the occluded part of the pupil. This does not offer any insight to the source of the problem and is a bit of a mystery itself.
(9 Oct 2013 GMB).