We received our first 4 cpu R820s, v2 processors. We have a lot of R810, running rhel and oel.
We put 2 of the new R820 in one datacenter, and 2 in the other. They will be production oracle database servers and standby replication servers.
All 4 servers have exhibited the same issues. During installation, which is via PXE boot and custom kickstart environment. This installation method has never had issues with the R810s, or any vmware virtual server installations.
The installation gets to a certain point and the server immediately reboots, throwing up an IERR on cpu0, followed by info about devices at bus 0 dev 1, bus 1 dev 0, or some iteration of that. There is also a message about pci express error sometimes.
At first, I thought we had hardware issues. A technician replaced pci network cards and updated firmware, etc, with no change in behaviour. I did finally get an OS laid down, OEL 6.2, on one of the servers, but when starting the OS, network initialization, specifically bnx2x drivers.
This led me to c
heck the support matrix for R820 with 4 v2 processors. I found the minimum is 6.5. So, my question is this:
Would trying to install something before 6.5, 6.2 in this case, cause this sort of behaviour, perhaps during a part of the install where it deals with pci hardware? Though there aren't progress logs of any of the installs to see where it bombed on each, the timings seems very close for each one, so I suspect it was at the same part of the installation each time.
I have successfully installed 6.5 on each server, with no issues, hardware or otherwise.