Re: Time jumping forward issue under OpenBSD 6.6 VMM
Hi Ian, I think this is a common problem, and has been an issue from time to time on KVM machines also, you can try an alternate counter and see how you get on, if you check out Mischa Peters talk on VMM he has some workarounds that he has deployed in production. but for an NTP server the accuracy of the clock may cause you too much problems.. as I said we had issues before with KVM and the timecounter hardware... but the newer timecounters in recent versions of OpenBSD have worked really well for us. I hope this helps, Tom Smyth On Tue, 3 Mar 2020 at 17:05, Ian Gregory wrote: > > On Tue, 3 Mar 2020 at 15:47, mabi wrote: > > It looks like there is a time issue on that VM although I am running the > > default ntpd of OpenBSD 6.6 and I have added the following parameter into > > my /etc/sysctl.conf on that VM: > > > > kern.timecounter.hardware=tsc > > I've had similar issues with timekeeping within guests of VMM, > although there are improvements in -current with the pvclock time > source. Since the fix below I now see occasional instances of the > clock stepping by a few whole seconds (typically less than 8s) but > it's much less frequent and the magnitude is within the bounds of what > ntpd can correct. > > See > http://openbsd-archive.7691.n7.nabble.com/pvclock-stability-tp376946p377922.html > for some backstory > > (aside: I see similar small occasional clock jumps of an integer > number of seconds on OpenBSD-6.6 guests using tsc running on a VMware > ESXi host) > > Regards > Ian > -- Kindest regards, Tom Smyth.
Re: Time jumping forward issue under OpenBSD 6.6 VMM
‐‐‐ Original Message ‐‐‐ On Tuesday, March 3, 2020 8:52 PM, Jordan Geoghegan wrote: > The clocks are basically broken on vmm. The pvclock stuff is definitely > an improvement, but it's still not there. There's still a ways to go > before we have proper, reasonably accurate clocks for vmm guests. Thanks Jordan for this info. For now I might just avoid VMM for time sensitive services and use bare metal with OpenBSD of course. I will for sure go back to VMM once the clock is more accurate.
Re: Time jumping forward issue under OpenBSD 6.6 VMM
On 2020-03-03 07:43, mabi wrote: Hello, I am running an OpenBSD 6.6 VMM server with a few virtual machines also running 6.6 and noticed that one of them which is running Dovecot 2.3.9 package outputs the following warning every few 10-15 seconds: dovecot: imap: Warning: Time jumped forwards 28.294585 seconds It looks like there is a time issue on that VM although I am running the default ntpd of OpenBSD 6.6 and I have added the following parameter into my /etc/sysctl.conf on that VM: kern.timecounter.hardware=tsc Is there anything else I can do to avoid this time issue in my VM? Best regards, Mabi The clocks are basically broken on vmm. The pvclock stuff is definitely an improvement, but it's still not there. There's still a ways to go before we have proper, reasonably accurate clocks for vmm guests.
Re: Time jumping forward issue under OpenBSD 6.6 VMM
‐‐‐ Original Message ‐‐‐ On Tuesday, March 3, 2020 5:58 PM, Ian Gregory wrote: > I've had similar issues with timekeeping within guests of VMM, > although there are improvements in -current with the pvclock time > source. Since the fix below I now see occasional instances of the > clock stepping by a few whole seconds (typically less than 8s) but > it's much less frequent and the magnitude is within the bounds of what > ntpd can correct. > > See > http://openbsd-archive.7691.n7.nabble.com/pvclock-stability-tp376946p377922.html > for some backstory > > (aside: I see similar small occasional clock jumps of an integer > number of seconds on OpenBSD-6.6 guests using tsc running on a VMware > ESXi host) Thank you Ian for your answer. Unfortunately I don't understand too much of that post you linked. But if I understand you correctly you are suggesting that I should change on my VM: kern.timecounter.hardware=pvclock and the drift or jumps should be better than with "tsc" ?
Re: Time jumping forward issue under OpenBSD 6.6 VMM
On Tue, 3 Mar 2020 at 15:47, mabi wrote: > It looks like there is a time issue on that VM although I am running the > default ntpd of OpenBSD 6.6 and I have added the following parameter into my > /etc/sysctl.conf on that VM: > > kern.timecounter.hardware=tsc I've had similar issues with timekeeping within guests of VMM, although there are improvements in -current with the pvclock time source. Since the fix below I now see occasional instances of the clock stepping by a few whole seconds (typically less than 8s) but it's much less frequent and the magnitude is within the bounds of what ntpd can correct. See http://openbsd-archive.7691.n7.nabble.com/pvclock-stability-tp376946p377922.html for some backstory (aside: I see similar small occasional clock jumps of an integer number of seconds on OpenBSD-6.6 guests using tsc running on a VMware ESXi host) Regards Ian
Time jumping forward issue under OpenBSD 6.6 VMM
Hello, I am running an OpenBSD 6.6 VMM server with a few virtual machines also running 6.6 and noticed that one of them which is running Dovecot 2.3.9 package outputs the following warning every few 10-15 seconds: dovecot: imap: Warning: Time jumped forwards 28.294585 seconds It looks like there is a time issue on that VM although I am running the default ntpd of OpenBSD 6.6 and I have added the following parameter into my /etc/sysctl.conf on that VM: kern.timecounter.hardware=tsc Is there anything else I can do to avoid this time issue in my VM? Best regards, Mabi