I did a test, point each miner to different pool and guess what happen after a few hours?About the miner restart phenomenon, please refer to FAQ
Why mining software automatically restart?
Mining software support self recovery from network fault and drive fault function, with watchdog monitoring for the whole system , when an anomaly is detected, the mining software will automatically restart, trying to self recovery from the fault, so mining software auto restart is a normal phenomenon.
General, restart phenomenon should be random, but if your local network break down, the miner at the same router may restart at the same time and try to recover the network link.
Of course, the remote pool bad status may cause the miner restart also.
Further more, if you config the temp monitor, the miner will shut down when the temp above the shutdown threshold, and you should restart it manually.
The nicehash extranonce bug already fixed at 3.5.6.
They restarts all in same time and all are connected to the same router so it looks like the network must mostly cause this restarts.