It cant be the router. On top of these s19 pro I also have 220 units of S17 running on the same network. So there are over 500 units on the network. I increase the network to be able to handle 750+ ip address. It goes from 10.1.10.1 - 10.1.12.255
Just because you changed your network settings to class A, it doesn't mean the infrastructure can handle all the traffic, this why some routers/switches cost 10x the price of another router/switch, but well since the miners completely shut down, I would rule out the network thing altogether.
I would like to hear more explanation on the "shut-down / turn off f" part, does the PSU itself shut down and its fan stop spinning or does the PSU shut down its main rail and keep the 12v running so the miner's fan and control board work but the hashboards don't?
Since you mentioned that you changed the control board and PSU, what I would do is move one of the bad miners to another rack, just to ensure that this isn't an electricity-related issue, if it is not, then chances are one or some of these boards have a loose heatsink which loses contact with the chip when the mine gets warm, this is very common with the 17 series and could be with the 19 series too since they gears are new, we don't have much info about them, to troubleshoot that you will need to run the miners with 1 board at the time to figure out which board/boards are causing the problem.
We have a business modem / router from comcast. It should handle all these miners just fine. Also we have good upload and download speed. So what happens --> at random times the miners turn off but to be more exact they look like they restart and dont start mining after restarting. Or they just stop mining. The kernel is showing that they are most likely restarting. Here is some kernel info from a miner that has been running for 7 hours:
2021-01-19 23:48:58 freq = 525, percent = 90, hcn = 12480, timeout = 449
2021-01-19 23:48:58 set_start_time_point total_tv_start_sys=121 total_tv_end_sys=122
2021-01-19 23:48:58 set_voltage_by_steps to 1307.
2021-01-19 23:54:31 set_voltage_by_steps to 1287.
You can see that its doing a == set_voltage_by_steps function -- it auto regulates the voltage as time goes on. The miners that have the restart issue are not showing multiple entrys for set_voltage_by_steps after they stop mining. That gives me the impression that the miners restarted since there is only one entry from the bad miner and once it restarted its not mining anymore. The problem could be the heat sinks but it seems that bitmain acknowledged this problem with the s17 series and supposedly fixed the issue on the S19. So i would not put much thought into the heat sink problem. I will try to diagnose the panel for these miners. I can also try moving a bad miner to a different location and see if the problem goes away. If anyone has any other thoughts let me know. This problem is really hard to figure out since there is ZERO kernel info.