I often have to observe during less-than-ideal conditions. Even with long exposures (~20sec) and 4.4 binning, plate solving is problematic.

The irritating behavior occurs thus:

Image captured
Plate-solving fails fairly quickly.
But before the failure, another image starts capture.
Then the plate solver (I am guessing here) tries to capture another image, but the camera is busy.
So it goes into wait mode, for 10 seconds IIRC.
Meanwhile, the busy capture may finish before, or after, the wait interval.
The result is a loop that takes a while to resolve, depending on the timing involved.
This sure looks like a thread or process starvation issue.

My suggestion: abort any camera captures as soon as the solver fails, to clear the error conditions.