I've just be forced to replace my NVIDIA motherboard with a ASRock which I know is not ideal, initially the system failed to load. I forced the BIOS to MPS 1.1, this didn't solve the problem. I therefore reinstalled ESX 3.0.1 with the NOFD NOAPIC NOUSB switches, and now the system loads, and the VM's run perfectly. On the ESX server console page displayed after boot (The one with "VMware ESX Server version 3.0.1" in the first line) there are the following two lines in red:
TSC: 324329654 cpu0:0)Chipset: 664: MPS found but host is using PIC
TSC: 324348328 cpu0:0)Chipset: 665: Make sure that if 'noapic' is used, it is on purpose
Can anyone explain what the 'MPS found but host is using PIC' means, and is there a way to remove or solve this problem.
I know why the NOAPIC message is displayed, but I would like to know what impact this may have on my system if any.
Regards
This may helpful to u
http://kb.vmware.com/selfservice/microsites/search.do?cmd=displayKC&externalId=1081
Thanks for the link, unfortunatley this refers to HP servers, ideally I'd like to understand what these messages mean so that I can hopefully resolve the situation.
Every get an anwser to this?
The messages go together, and are simply pointing out that you are using the "noapic" option on a system that appears to have APICs.
Using "noapic" causes all interrupts to be handled by the BSP (pcpu 0). This has obvious scalability issues, but should not affect stability or correctness. Servers on the HCL should rarely, if ever require the "noapic" option.
Can I turn off the NOAPIC option then? How would I do that?
Thanks for the info you sound like a total guru! Can I turn off the
NOAPIC that is apparently enabled? Or is that a bad thing to do...
I appreciate your help thanks for your time
Michael
Whether or not you can turn off the noapic option depends on the reason why you turned it on.
To turn off "noapic", you need to remove the option from the "kernel" lines in /boot/grub/grub.conf. At first, I would recommend removing it only from the "VMware ESX Server" configuration, and leaving it in the debug mode and troubleshoot mode. That way if the server fails to boot, you can reboot into troubleshoot mode and put it back.