2012-03-16, 15:23
I don't think that the preclear is cpu intensive at all. If anything it is going to stress your SATA controller/chipset the most. And if you recall the unraid hardware compatability wiki said to stay away from the nvidia chipset you have. One of the best stress tests you can do for an unraid machine is to preclear multiple drives at the same time - the more the better. You already tried this and I'd say that it did it's job in pointing out that you have an issue.
FWIW, I had an issue similar to yours the last time I built a server. The drives would preclear just fine but I would get random lockups of the server. I ran memtest for three days and got no errors but I swapped the RAM to another machine (my HTPC) and I got a blue screen on it within a day. So bad RAM may not always show errors in memtest. At this point I think your best bet is to swap out the RAM for another set regardless of whether or not it passes memtest. If you can swap it with another set that is already working in a machine you already have then that is ideal. If the RAM does not correct it then try the PSU (again swapping it with a known good one from another machine if possible), then the MB and the last thing I'd change is the CPU.
FWIW, I had an issue similar to yours the last time I built a server. The drives would preclear just fine but I would get random lockups of the server. I ran memtest for three days and got no errors but I swapped the RAM to another machine (my HTPC) and I got a blue screen on it within a day. So bad RAM may not always show errors in memtest. At this point I think your best bet is to swap out the RAM for another set regardless of whether or not it passes memtest. If you can swap it with another set that is already working in a machine you already have then that is ideal. If the RAM does not correct it then try the PSU (again swapping it with a known good one from another machine if possible), then the MB and the last thing I'd change is the CPU.
