[ntp:questions] Suppressing IP6; Keeping /dev/ppsX; simple MSF clocks and monitors

Chris Hastie chris at nospam.oak-wood.co.uk
Thu Nov 20 08:45:26 UTC 2003


A recent need to reboot has brought to light a couple of problems in my
configuration:

IP6
===
I don't use it, don't know much about and doubt my ISP supports it. I
recently added an additional server to my ntp configuration, at the time
updating ntp.conf but rather than restarting ntpd adding it using ntpdc.
This worked fine. But when I actually came to have to restart ntpd the
host was resolved to an IP6 address and ntpd started complaining that
there was no route to host. The only way I've managed to start ntpd and
include this host is to specify it by IP4 IP address in ntp.conf

Maybe this is an OS / resolver config problem rather than ntpd. I'm
using 4.2.0 on FreeBSD 5.1

/dev/ppsX
=========
I'd soft linked /dev/pps1 to /dev/ttyd0, but this link didn't survive
the reboot, resulting in ntpd complaining about the lack of /dev/pps1.
Should it have survived the reboot? If not is there a way to ensure it
does? If not, what's the sensible way around this - edit
/usr/local/rc.d/ntpd.sh to create it just before ntpd starts?

Simple MSF clocks and monitors
==============================
I've noticed after producing a few graphs that my server synchronised to
a simple MSF clock (the Jonathan Buzzard one) tends to be pretty close
to other servers at times that I'm generally sat at the console but have
offsets of around -1.6ms at other times. I couldn't see that I was
somehow changing the load significantly, and there were odd times when
the me-at-the-console sync continued all day although I wasn't at the
console. After some pondering I hypothesised that sync was closer when
the monitor was turned on. After leaving it on all day yesterday this
does appear to be the case. Could anyone suggest a mechanism for this?
Is it likely to be a simple as the magnetic field from the monitor
affecting the radio clock such that re-positioning the radio clock
further from the monitor will produce more consistent time? Or is the
issue more likely to be in proximity of cables or some obscure thing
inside the PC box?
-- 
Chris Hastie



More information about the questions mailing list