System clock speedup with latest kernels
William Lovaton
walovaton at yahoo.com.mx
Sun Mar 12 15:36:02 UTC 2006
Hi,
El vie, 10-03-2006 a las 10:53 -0700, Stanton Finley escribió:
> On Fri, 2006-03-10 at 12:39 -0500, John Thacker wrote:
> > Anyone else seeing this:
> > https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=184593 ?
> >
> > With the last two kernels, 2.6.15-1.2032_FC5 and 2.6.15-1.2038_FC5,
> > the system clock starts running too fast for me. It gains about an
> > hour per day. It runs so fast that NTP won't sync to any clock
> > servers because the jitter grows far too fast. Rebooting to
> > kernel-2.6.15-1.2009.4.2_FC5 works perfectly. I can switch back and
> > forth between the working kernel-2.6.15-1.2009.4.2_FC5 and the two
> > broken kernels, and the behavior shifts between working and running
> > too fast each time. (The hardware clock runs perfectly fine too.)
> >
> > Anyone else seeing it? Any more information that would be useful?
> > I assume this is related to the "timer fixes" from git9 mentioned
> > in the 2028 update, although frankly I don't know enough to be sure.
> > Obviously, this is a blocker for me personally, but I haven't seen
> > any other complaints.
> >
> > John Thacker
> > --
> > fedora-devel-list mailing list
> > fedora-devel-list at redhat.com
> > https://www.redhat.com/mailman/listinfo/fedora-devel-list
>
> I have the same issue. Try adding "noapic nolapic" to the end of the
> "kernel" line in /boot/grub/grub.conf.
I have two IBM xSeries 445 servers running a web server and a database,
they are SPM 8X systems with hyperthreading. And every 2 or 3 days they
get out of sync from each other and from the real world time. Is there
any regression if I disable apic and lapic in the kernel parameters?
They are under heavy load usually.
What they are for (apic and lapic)?
-William
__________________________________________________
Correo Yahoo!
Espacio para todos tus mensajes, antivirus y antispam ¡gratis!
Regístrate ya - http://correo.yahoo.com.mx/
More information about the devel
mailing list