[ntp:questions] Linux vs. ntpd 4.2.4p4

Heiko Gerstung heiko.gerstung at meinberg.de
Wed Jan 16 08:30:56 UTC 2008


I observed a strange behavior of the Linux kernel which I 
patched with Rodolfo's LinuxPPS in order to support the PPS of my 
Meinberg GPS receiver.

My first tests showed that this version of ntp seems to converge very 
slooooow despite an initial offset of +/- 500us. It seems that ntp's 
frequency adjustments are not "aggressive" enough, because the offset 
grows from 0.5ms to 10ms after ~ 2 hours (with deceasing speed) and then 
it turns around and takes another 2-3 hours before it reaches our 
beloved zero line.

I think there was some discussion on the hackers list in the last days 
in regards to PLL constants, this might be a related issue.

In order to get it to correct the clock / frequency faster, I changed my 
kernel settings from HZ=250 to HZ=1000 and voila, it looks much better. 
Conversion is faster and with a correct drift file takes only ~10-15 
minutes until the offset reaches my "target area" of +/- 5us .

A strange side effect keeps me from using this kernel setting: When I 
plug in a USB stick (without doing anything with it, no mounting or 
anything else)  and remove it , ntp immediately reports an offset of 
10ms (or sometimes even 40ms). IMHO this indicates time jumps caused by 
lost timer ticks, but I am just wild guessing here. At least it is fully 
reproducible and after plugging the USB stick in and out for a while 
(ignoring the ridiculous comments from my co-workers), the offset grows 
to 150ms and more, depending on the number of times I removed the USB stick.

This definitely is a kernel issue and has nothing to do with ntp, but I 
would like to know if someone else has the chance to try this out on a 
machine where she/he can modify th kernel configuration.

I will do some further testing before I report a (kernel) bug, but I 
would be happy to get some more results before doing that.

Best Regards,

More information about the questions mailing list