We perform probe-tack experiments on highly viscous silicone oils. Direct observation during the experiment shows the existence of several mechanisms for releasing the stress. Beyond bubble nucleation and instantaneous growth ( reported in a previous work), delayed bubble growth is observed at higher traction velocities. At even higher velocities, cracks at the interface between the plate and the liquid appear before the bubbles have grown to their full size. Bubbles and cracks are thus observed concomitantly. At much higher velocities, cracks develop fully before the bubbles can be seen. We present a theoretical model that describes these regimes, using a Maxwell fluid as a model for the actual fluid. The predictions for the force peak are in qualitative agreement with the data. We discuss to what extent a Maxwell model is adapted to this situation. In particular, we discuss the threshold value for cavitation or cracking in the case of a purely elastic material and obtain a complementary, tentative interpretation of Dahlquist's criterion for stickiness.