virtualbox.org

I am running VirtualBox 2.1.2 on Vista 64bit with a RedHat 5 Update 2 64-bit guest and have noticed that the time keeps moving forward. I am not having the same problem with my 32-bit XP guest. Does anyone know wht the problem is and if there is a fix or is this a known bug?


Thanks!

If you install the GA, the Guest will sync it's time every few minutes with the host.

What is GA?

Short for Guest Additions.

GA don't help. In the default configuration (10 seconds between time syncs - who's 10 seconds btw?!) the clock inside guest fluctuates with tens of seconds back and forth.

Starting vboxadd-timesync with --interval 1 helps, but it's still a few seconds wrong. Now, 0.1 is much better, the clock is just ~10-20 seconds behind, but keeping almost a straight flow.

Almost is the key word. In fact the time still fluctuates, one second back corrections, and this still causes "make" for example to crash. And since we're using the virtual machines to build software, it's a complete disaster. They never make it to finish compiling all the software.

I'm terribly disappointed by this behavior, and the fact that there is no reaction yet to this bug being reported.

Try disabling all other time related services, ntp ...

I've tried all the possible combinations, ntpd or no ntpd, vboxadd-timesync or not ... The guest just runs too fast, all the animations are too fast, keystrokes show maaaaaaaany times the same letter if the hit is not brief enough, it's just hooked on another time line

However, this seems to be related to older kernels. I don't have yet enough statistics for this, but an OpenSuSE 11.1 (2.6.27.7-9) runs just fine, while two RH-based distributions with 2.6.9 and 2.6.18 show this weird race.

costing wrote:I've tried all the possible combinations, ntpd or no ntpd, vboxadd-timesync or not ... The guest just runs too fast, all the animations are too fast, keystrokes show maaaaaaaany times the same letter if the hit is not brief enough, it's just hooked on another time line

However, this seems to be related to older kernels. I don't have yet enough statistics for this, but an OpenSuSE 11.1 (2.6.27.7-9) runs just fine, while two RH-based distributions with 2.6.9 and 2.6.18 show this weird race.

You could try adding kernel options 'noapic nolapic nosmp' to the guest kernel command line. Those together with use of ntpd solved time issues for me with VMWare.

Thanks TrevorPH! Adding the kernel options fixed the problem on older kernels.