I have been having all sorts of issues getting a second mining rig setup. The pc is very unstable and can run from anywhere between a few minutes and a few hours before it dies. When it dies the machine floods the network with Pause packets that takes out anything else on the same switch as detailed here. As the machine is headless and I can only access it via SSH I can't get any information from the machine about what the problem is.
In trying to diagnose this problem I have swapped or replaced every single component: processor, power supply, power cables, memory. I have also thoroughly tested all GPUs on another machine before moving them over.
In my last test I had miner B running in my second case and miner C running without a case. Miner C runs fine with no issues or instability at all. Miner B crashes regularly.
I took the motherboard (including processor, memory and hdd) from miner C and swapped it with the motherboard from miner B.
Now miner C is crashing and miner B is fine.
As I said, I have swapped different power supplies and cables already.
I now suspect that the "case" is somehow responsible for this.
I say case.. The case is actually just a frame with shelves for GPUs. The motherboard is screwed to a sheet of plastic.
Can anyone suggest what is going on and what I can try next?
All 3 machines have identical motherboards and have identical Ubuntu server OS installs. Machine A runs with 12 GPUs with no stability issues at all.