ntpd on Red Hat

[Kent Borg]

On Thu, Apr 04, 2002 at 01:29:20PM -0500, Jerry Feldman wrote:
> It appears not to be taking the external servers. In contrast, I've 
> included the same table from my home system below. (Note that at home I'm 
> just using some secondary time servers). 
> 
> --------------------- BLU Server --------
> [root at asgard gaf]# /usr/sbin/ntpq -p
>  
>     remote           refid      st t when poll reach   delay   offset  
> jitter
> ============================================================================
> ==
> *LOCAL(0)        LOCAL(0)        10 l   19   64  377    0.000    0.000   
> 0.000
>  jj.cs.umb.edu   0.0.0.0         16 u    -   64    0    0.000    0.000 
> 4000.00
>  ourconcord.net  0.0.0.0         16 u    -   64    0    0.000    0.000 
> 4000.00
>  NAVOBS1.MIT.EDU 0.0.0.0         16 u    -   64    0    0.000    0.000 
> 4000.00
>  sirius.ctr.colu 0.0.0.0         16 u    -   64    0    0.000    0.000 
> 4000.00
>  mead.harvard.ed 0.0.0.0         16 u    -   64    0    0.000    0.000 
> 4000.00


I say network problem, someone is blocking ntp trafffic, but ntpd
itself is running.  In the above output ntpd managed to talk to itself
19 seconds ago, will try again in 45 seconds, and it was successful
the last 8-times it tried to talk to itself.  Not conclusive, but it
passes a loopback-level sanity test.

A fairly generic (non-NATing) firewall where I work doesn't let my
notebook talk to external ntp servers.  I have not looked at the
details of the ntp protocol, but I can imagine that the delicate
requirements for bouncing data back and forth to estimate timings
might easily get blocked by a firewall.

Also, before I got my basement server working as an NTP server I had a
hard time getting ntp service to my notebook at all.  When at home I
couldn't get time from my favorite external servers sucessfully, I
think it was because my basement server was talking to the same
servers as the notebook and the protocol likely preserves some state
info that got confused by how one IP address had two different
concepts of time.  At least that was my guess; as I said, I have not
learned the protocol.

Do you have port 123 open for both UDP and TCP traffic?  Have you run
successfully a Linux NTP client on this network before?


-kb, the Kent who wants to understand this one.