Hi Constantino,
Your problem is a direct result of very poor drilling. The 'cobblestone'
appearance of the plating is a direct reflection of the underlying rough hole wall
with protruding glass bundles, and glass bundles ripped out of the epoxy
matrix. These capillary gaps between the glass bundles and the resin allow plating
into these spaces--causing possible CAF shorting--but more importantly rob Cu
from plating a uniform hole wall. This creates this 'cobblestone' appearance
which is only the visual clue of significant stress risers in the PTH hole
wall. I would suspect that the second plating run was done because bare board
testing showed open PTH structures. I would also suspect you have cracked PTH
barrels somewhere in these PCBs--you just have not found them yet.

Werner