Smilin - indeed wires and connectors can fail. I've had it happen to me. It's usually when accompanied by "sharp" high frequency vibration. (The extreme case was in a vibration test environment when the tech pulled a wire from the table's drive while it was still powered up. 100G shocks do things to even small electronics.)
I'm troubled by the coincidence when BEA says it is also triggered by good data transfer of bad data as their working point. It may be wiring. I don't see that as a sure thing. It seems to push coincidence too hard. And I don't see anything that makes it a sure thing rather than as a path for conjecture. It's surely a valid point for conjecture. I'd not take it as the only path for conjecture.
Edit: I think we agree. I am in disagreement with A33Zab's assurance this is the problem.