[ntp:questions] Re: ntp server 'rejected'

Ronan Flood ronan at noc.ulcc.ac.uk
Fri Apr 21 23:46:41 UTC 2006


I wrote:

> I think the real problem lies with the Galleon server's strange
> dispersion figures; anything else is only working around that.

I think I've sussed this: the refclock precision is actually set
to *zero*, which means the filter dispersion is approx 2^0 == 1.

That looks like a bug in whatever is feeding data to refclock_shm,
and possibly also an oversight in refclock_shm not checking that
it has a sensible precision value before passing it upwards.

This also exposes a long-standing bug in ntpq.c\decodeint(), where
"&val" is used (twice) instead of "val", which screws-up display
of a zero precision value; eg:

ntpq> rv &2
status=9624 reach, conf, sel_sys.peer, 2 events, event_reach,
srcadr=127.127.28.1, srcport=123, dstadr=127.0.0.1, dstport=123,
leap=00, stratum=0, precision=10, rootdelay=0.000, rootdispersion=0.000,
[...]
ntpq> raw
Output set to raw
ntpq> rv &2
status=0x9624,
srcadr=127.127.28.1, srcport=123, dstadr=127.0.0.1, dstport=123, leap=0,
stratum=0, precision=0, rootdelay=0.000, rootdispersion=0.000,
[...]

Compare decodeint() with decodeuint() ...

-- 
                      Ronan Flood <R.Flood at noc.ulcc.ac.uk>
                        working for but not speaking for
             Network Services, University of London Computer Centre
     (which means: don't bother ULCC if I've said something you don't like)




More information about the questions mailing list