Guest time sync busted in 3.0.6?
-
rfm
- Posts: 18
- Joined: 8. Sep 2008, 20:45
- Primary OS: MS Windows 10
- VBox Version: VirtualBox+Oracle ExtPack
- Guest OSses: Ubuntu, Windows
- Location: Palo Alto, California, USA
Guest time sync busted in 3.0.6?
I have a Ubuntu 9.04 guest on a Ubuntu 9.04 host; it runs my mail system, including dovecot. I upgraded to vbox 3.0.6 today, upgraded the guest to 3.0.6 host additions; now every few hours the guest clock is getting set backwards by an hour or so. This causes dovecot to crash. Is the guest time sync somehow damaged in 3.0.6? Can I work around this by taking /etc/init.d/vbox-services out of the run level?
-
rfm
- Posts: 18
- Joined: 8. Sep 2008, 20:45
- Primary OS: MS Windows 10
- VBox Version: VirtualBox+Oracle ExtPack
- Guest OSses: Ubuntu, Windows
- Location: Palo Alto, California, USA
Re: Guest time sync busted in 3.0.6?
Looking at the logs, both times this has happened the clock went back by 2147 seconds (35 minutes 47 seconds.) Weird.
-
rfm
- Posts: 18
- Joined: 8. Sep 2008, 20:45
- Primary OS: MS Windows 10
- VBox Version: VirtualBox+Oracle ExtPack
- Guest OSses: Ubuntu, Windows
- Location: Palo Alto, California, USA
Re: Guest time sync busted in 3.0.6?
Thinking about it, 2147 seconds would be 2^31 nanoseconds rounded to seconds. That almost has to mean something.
I now see there are other people having clock problems on earlier versions; I never saw it until 3.0.6 and then I saw it immediately.
Last night when I posted I noticed the guest clock was running quite a bit faster than the host; overnight I didn't get any of the giant backward adjustments, and the clock has only run ahead about 45 seconds. I'm now logging the clock skew on the two systems to see if I can spot a trend.
I now see there are other people having clock problems on earlier versions; I never saw it until 3.0.6 and then I saw it immediately.
Last night when I posted I noticed the guest clock was running quite a bit faster than the host; overnight I didn't get any of the giant backward adjustments, and the clock has only run ahead about 45 seconds. I'm now logging the clock skew on the two systems to see if I can spot a trend.