I am sorry to hear that the problem is not so simple and easily determined/solved.
I admit I'm not a Linux 'internals' expert. (Unfortunately you may not find one here, but I will cede the case to anyone who things they can help with it.)
One last observation: A long time ago (in the late 1980s) I formed a principle when working with PCs. It runs something like this:
- if the problem is regular, repeats itself *exactly*, and without fail... i.e. "if same problem in same circumstances 100 % of the time"... then it is a problem in the software.
- if the problem is irregular, not always repeating in the same circumstances... then it is probably hardware at fault.
This is based on the notion that incorrectly configured software is always operant a certain way (even if defectively), but hardware... can be flakey.
(Of course, if the variation in circumstances is invisible to one's examination, then variation in problems will occur... so I present this with a strong caveat. As a principle, it can be wrong.)
But please consider this advice. It might help you see what is wrong with the circumstances...? it's the last help I can offer on this, I'm sorry to say.
I have Opted Out, opening the case to other Experts in this category.