topic Re: Problem in ntptrace from NTP client machine in Operating System - HP-UX

Problem in ntptrace from NTP client machine

Balaji_6 — Tue, 04 Feb 2003 11:04:38 GMT

Hi Admins,

I am planning to configure NTP server in our office and is going to act as a server for 100 Unix machines.
Our Primary NTP Server is far away from our office and I am synchronizing our NTP server with Primary NTP server through WAN.
I did following setup in our
NTP Server,
Entry added in /etc/ntp.conf file.
server 127.127.1.1

fudge 127.127.1.1 stratum 10 # show poor quality

broadcast 15.17.176.255

Then, following entry are added in /etc/rc.config.d/netdaemons file,
export NTPDATE_SERVER=PRIMARYSERVER.hp.com
export XNTPD=1

Then i synchronized our NTP server date by the command,
ntpdate PRIMARYSERVER.hp.com
and restarted the xntpd service.

In the client side, i added following entries,

In client side,i configured the following setup in /etc/rc.config.d/netdaemons.
export NTPDATE_SERVER=OURNTPSERVER.hp.com

export XNTPD=1

By the command ntpdate, i synchronized the clock.
ntpdate OURNTPSERVER.hp.com
And i started the xntpd service.

Now the problem in our setup is, if i tried to trace the ntp server from my client, its giving following error and stops at our NTP server giving error as 'Timeout'
OURNTPSERVER.hp.com: stratum 11, offset 0.038064, synch distance 0.01022
127.127.1.1: *Timeout*
The same thing i did from my NTP server, to trace our Primary NTP server, its reaching without having any error as given below,
PRIMARYSERVER.hp.com: stratum 2, offset -0.019422, synch distance 0.17390
SERVER.HP.COM: stratum 1, offset -0.013017, synch distance 0.00000, refid 'GPS'

As per the error i feel our client machines are start trying from stratum 11 only and not from stratum 3, so i dont know how to configure these setup in NTP server.

Can you pleases correct me if i done any misconfiguration in our NTP server.

-Balaji

Re: Problem in ntptrace from NTP client machine

David_246 — Tue, 04 Feb 2003 11:15:10 GMT

Hi,

Be aware that most routers (you're talking about a WAN) block the ntp broadcast. Just change the setup for each client to say :

server ntpserver

Make sure you add the alias ntpserver to your current ntpservername (you can switch ntpserver any time you want now, without editing 100 clients).

In this case you don't have a broadcast, but a direct connect, this lowers the network traffic and, still keeps it flexible.

Regs David

Re: Problem in ntptrace from NTP client machine

David_246 — Tue, 04 Feb 2003 11:19:33 GMT

Hi,

Of course you have to add the alias ntpserver in DNS or NIS :)

The real ntpserver, should have only it's local clock or remote ntp server in the ntp.conf

Regs David

Re: Problem in ntptrace from NTP client machine

Ron Kinner — Tue, 04 Feb 2003 16:14:43 GMT

Perhaps I'm missing something but shouldn't you have your real server listed in
/etc/ntp.conf
of your server?
by only having the local clock you have not given it much choice of what to advertise.

Also since you are trying to broadcast to everyone on the 15.17.176 shouldn't your clients just have the line:

broadcastclient

in /etc/ntp.conf?

Ron

Re: Problem in ntptrace from NTP client machine

Balaji_6 — Mon, 10 Feb 2003 16:09:35 GMT

Hi,

Thanks for your value suggestions,
I just added our real server OURNTPSERVER.hp.com to /etc/ntp.conf file and /etc/rc.config.d/netdaemons and then i traced the same its reaching to STRATUM 1 server without have any errors.

Once again thanks Ron & David.

-Balaji