OpenNTPd howto?

Tue Jul 1 19:10:32 UTC 2008

On Jul 1, 2008, at 1:21 PM, Bill Moran wrote:

> In response to B. Cook <bcook at poughkeepsieschools.org>:
>
>> Hello All,
>>
>> Not sure what I am missing, but I am.
>>
>> so I put openntpd on a machine (10.20.0.16)
>>
>> cat ntpd.conf | egrep -v ^#
>>
>> listen on 0.0.0.0
>> server clock.nyc.he.net
>>
>> then start it and it looks like it does:
>>
>> USER     COMMAND    PID   FD PROTO  LOCAL ADDRESS         FOREIGN
>> ADDRESS
>> _ntp     ntpd       15751 4  udp4   10.20.0.16:55180
>> 209.51.161.238:123
>> _ntp     ntpd       15751 6  udp4   *:123                 *:*
>>
>>
>> Strange thing one:
>>
>> root at core [/usr/local/etc]# 30 > ntpdate -b clock.nyc.he.net
>>  1 Jul 12:43:52 ntpdate[48881]: the NTP socket is in use, exiting
>>
>> root at core [/usr/local/etc]# 31 > /usr/local/etc/rc.d/openntpd stop
>> Stopping openntpd.
>>
>> root at core [/usr/local/etc]# 32 > ntpdate -b clock.nyc.he.net
>>  1 Jul 12:49:57 ntpdate[70917]: step time server 209.51.161.238
>> offset 358.732506 sec
>>
>> Why when it was running did it not update the clock on the server?
>
> It was working on it.  You should read up on NTP a bit so you  
> understand
> how it works.  NTP does not "set" the clock unless you explicitly tell
> it to (I believe the -s switch in openntpd).  Instead, it speeds up or
> slows down the clock to bring it into adjustment, which prevents  
> software
> from seeing a sudden and space-time fabric-ripping shift in time.
>
> If you let openntpd run for a while, possibly a few hours, you'd see  
> the
> time come in to sync.
>
>> From a different computer I can not get the time from the server
>> running openntpd.
>
> What error do you get?  Run ntpdate -d on the other computer to see  
> _why_
> it's refusing to sync.  I would guess it's because the OpenNTPd server
> knows that it's not in sync yet, and thus refuses to sync other  
> machines.
>
> -- 
> Bill Moran
> http://www.potentialtech.com

Thanks for the clue to the answer.

Here is the output:

pmsbsdsrv# ntpdate -d 10.20.0.16
  1 Jul 13:31:00 ntpdate[899]: ntpdate 4.2.0-a Sun Feb 24 16:32:49 UTC  
2008 (1)
transmit(10.20.0.16)
receive(10.20.0.16)
transmit(10.20.0.16)
receive(10.20.0.16)
transmit(10.20.0.16)
receive(10.20.0.16)
transmit(10.20.0.16)
receive(10.20.0.16)
transmit(10.20.0.16)
10.20.0.16: Server dropped: strata too high
server 10.20.0.16, port 123
stratum 16, precision -21, leap 11, trust 000
refid [10.20.0.16], delay 0.02599, dispersion 0.00000
transmitted 4, in filter 4
reference time:    00000000.00000000  Thu, Feb  7 2036  1:28:16.000
originate timestamp: cc14e855.037077ff  Tue, Jul  1 2008 13:31:01.013
transmit timestamp:  cc14e855.14ea3cc5  Tue, Jul  1 2008 13:31:01.081
filter delay:  0.02605  0.02600  0.02599  0.02599
          0.00000  0.00000  0.00000  0.00000
filter offset: -0.06838 -0.06845 -0.06845 -0.06845
          0.000000 0.000000 0.000000 0.000000
delay 0.02599, dispersion 0.00000
offset -0.068452

  1 Jul 13:31:01 ntpdate[899]: no server suitable for synchronization  
found

What I would like to have is a time server that works like how I think  
it works.  this 10.20.0.16 machine was updated and rebooted, and I was  
installing two new machines today and saw it wasn't syncing..

Is there a way to make a time server serve the time of the local  
computer, and then every hour update the server from a time server?   
Or just serve the time as soon as the server is enabled?

On the server I have done this:

# 30 > /usr/local/sbin/ntpd -s -d -f /usr/local/etc/ntpd.conf
listening on 10.20.0.16
ntp engine ready
reply from 209.51.161.238: offset 0.005419 delay 0.016668, next query 6s
reply from 209.51.161.238: offset 0.005236 delay 0.016233, next query 6s
reply from 209.51.161.238: offset 0.005288 delay 0.015782, next query 9s
peer 209.51.161.238 now valid
reply from 209.51.161.238: offset 0.005271 delay 0.016006, next query 9s
reply from 209.51.161.238: offset 0.005550 delay 0.015967, next query 7s
reply from 209.51.161.238: offset 0.005616 delay 0.016308, next query 7s
reply from 209.51.161.238: offset 0.005714 delay 0.015999, next query  
30s
reply from 209.51.161.238: offset 0.005995 delay 0.016138, next query  
32s
adjusting local clock by 0.005288s

but the client still sees this:

# ntpdate -d 10.20.0.16
  1 Jul 15:09:14 ntpdate[1105]: ntpdate 4.2.0-a Sun Feb 24 16:32:49  
UTC 2008 (1)
transmit(10.20.0.16)
receive(10.20.0.16)
transmit(10.20.0.16)
receive(10.20.0.16)
transmit(10.20.0.16)
receive(10.20.0.16)
transmit(10.20.0.16)
receive(10.20.0.16)
transmit(10.20.0.16)
10.20.0.16: Server dropped: Leap not in sync
server 10.20.0.16, port 123
stratum 2, precision -21, leap 11, trust 000
refid [10.20.0.16], delay 0.02599, dispersion 0.00000
transmitted 4, in filter 4
reference time:    cc14feea.d26147ff  Tue, Jul  1 2008 15:07:22.821
originate timestamp: cc14ff5a.7657d7ff  Tue, Jul  1 2008 15:09:14.462
transmit timestamp:  cc14ff5a.9fabbfcc  Tue, Jul  1 2008 15:09:14.623
filter delay:  0.02602  0.02600  0.02599  0.02599
          0.00000  0.00000  0.00000  0.00000
filter offset: -0.16169 -0.16162 -0.16163 -0.16162
          0.000000 0.000000 0.000000 0.000000
delay 0.02599, dispersion 0.00000
offset -0.161631

  1 Jul 15:09:14 ntpdate[1105]: no server suitable for synchronization  
found

it looks different/closer.. but clients still can not sync to it..

suggestions?