Hi Kristian, Clock is still absolutely synchronous, but ptree error already persists. Other clocksources than kvm-clock have significant jitter.
Unfortunally sks needs several peerings to get a reproduction of that problem. Christian Am 17.03.2012 18:50, schrieb Kristian Fiskerstrand: > On 16.03.2012 18:34, Christian Felsing wrote: >> Hi, > > > Hi Christian > >> same problem here. > > > ... > > >>> Anyhow, I'd check /sys/devices/system/clocksource/clocksource0/ >>> where I'd expect to see the files (i) available_clocksource, and >>> (ii) current_clocksource. > >> # cat >> /sys/devices/system/clocksource/clocksource0/available_clocksource >> kvm-clock tsc hpet acpi_pm > >> # cat >> /sys/devices/system/clocksource/clocksource0/current_clocksource >> kvm-clock > > Obviously I'm not familiar with the nature of this system, most > notably, whether this is a generic production system or an isolated > system for sks that you can play around with a bit more. The latter > would suggest that you could try changing the clocksource to see if > that helps with the stability. > > You can change the clocksource by setting the clocksource= kernel > option in your respective GRUB/LILO* setup, e.g. "clocksource=tsc" ** . > > * For grub it'd probably be in /boot/grub/menu.lst > > ** Please note that tsc in itself can be somewhat unstable e.g. if you > have CPU frequency scaling or similar, but it should have a watchdog > to monitor for that. I'm using it here as an example, without having > any specific recommendation for your individual system. > > > > > _______________________________________________ > Sks-devel mailing list > Sks-devel@nongnu.org > https://lists.nongnu.org/mailman/listinfo/sks-devel _______________________________________________ Sks-devel mailing list Sks-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/sks-devel