NetBSD-Bugs archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: kern/53016: Clock not stable

Michael van Elst writes:
>  3ms / minute is a rather usual value for a PC clock. That's an error
>  of about 50ppm or 4 seconds per day.

It is actually worse than that. When I stopped ntpd to check whether
this issue was fixed by latest kernel and forgot to restart it, the
clock was around 20 seconds off after about 10 hours or so. 

>  If ntpd cannot correct this, I'd guess there is a problem with the NTP
>  setup.

ntp can keep it inside few seconds, but as the poll interval can go to
quite big value for ntp, it can get several seconds from real time
before ntp fixes it. 

Also the clock drift is not completely stable, it seems to jump around
every now and then or something. When running the tests it seemed to
be quite stable few ms per minue for few minute tests we run, but
during few hours you have much bigger jumps in there.

For example now the clock offset was 1.7 seconds before the ntpd fixed

vielako (0:29) ~>ntpdate -q -u
server, stratum 2, offset 1.788493, delay 0.02718
 4 Apr 00:29:24 ntpdate[17818]: step time server offset 1.788493 sec
vielako (0:29) ~>

... (I did not do anything to fix the clock, but ntpd is running on background)

vielako (0:31) ~>ntpdate -q -u  
server, stratum 2, offset 0.000574, delay 0.02744
 4 Apr 00:31:39 ntpdate[10719]: adjust time server offset 0.000574 sec
vielako (0:31) ~>

Now one hour later the drift is already 2.7 seconds again.
vielako (1:39) ~>ntpdate -q -u
server, stratum 2, offset 2.789676, delay 0.02701
 4 Apr 01:39:45 ntpdate[16936]: step time server offset 2.789676 sec
vielako (1:39) ~>

Home | Main Index | Thread Index | Old Index