On the guest OS, are you using grub or lilo to boot the server?
I know on Suse Linux I had the same problem and even on reverse when the time ran ahead of the server (by hours in a single day).
Try adding the following boot parameter to the boot config file (Grub is /boot/grub/menu.lst, and Lilo should be /etc/lilo.conf):
Let me know how you get on...
Thanks for the fast replies, great forum!
I read the KB article and the white paper and will test out the various options, though none of them seem to be helpful.
The boot parameters, they have to be made inside of the VM, or on the host?
I tried it inside the VM, doesn't help much. Doing it on the hosts helps a little bit better, but is not perfect. The ping timing still is painfully wrong.
>The boot parameters, they have to be made inside of the VM, or on the host?
From the KB:
In a one-CPU virtual machine, add the following kernel command line parameters to the
nosmp noapic nolapic
there's also the following pdf available on timekeeping:
thanks for the fast reply. Seems that no options work so far. I will continue and try them out, but if you have any other idea, let me know.
Thanks for the tip, but adding
host.cpukHz = "2400006"
host.noTSC = "TRUE"
ptsc.noTSC = "TRUE"
did not help.
I recompiled the guest's kernel with 100 Hz, which did work so far. Not perfect, but it did work.
However, this is no solution to recompile all guest's kernels.
Anyone has an idea?
did you make sure all the power saving feature of your cpu are turned off in the machine bios... ?
As soon as I have physical access to the machine again, I will check that out. If that isn't the culprit, I'm out of luck.
Thanks for your help!
Neither cpuspeed nor powernow are loaded on host or guest.
This is driving me nuts... it works so flawlessly on the Core2 Duo machine.
Hope I can find a solution in the host's BIOS.
Thanks to all your help so far!
Disable C1E if you must.
I found something interesting in the hosts logs:
"rtc: lost some interrupts at 512Hz"