Bitmain hardware will lock up no matter the pool, it's just how they are.  They all of a sudden will "go out to lunch" and need a hard power cycle to kick them back into gear.

I had a large number of S5's deployed, I'd cycle a handful of them each week, and not always the same ones.  I just deployed 3x S4+ boxes last week.  Friday 2 of them went out, Sunday evening the other went out.  Power cycled today, back hashing, but just waiting for the next time.  Brand new.