[j-nsp] NTP service

Josef Buchsteiner josefb@juniper.net
Tue, 03 Sep 2002 18:22:15 +0200


At 05:38 PM 9/3/2002, Luis Eduardo D=EDaz Zazo wrote:
>Hello:
>
>I think the problem does exist. Although Asim and Yi-Kuo suggest the reason
>is that 128 seconds difference, if I understood Javier, he already took=
 that
>in mind and made himself sure that the difference didn't exist.
>
>I tried it in my company's bench and found the same problem as Javier did.
>
>I think the problem is as follows: while trying to synchronize a juniper to
>another juniper, the server always answers with an NTP response packet with
>its reference clock field set to zero, according to the packet analysis I
>did. So the client doesn't accept it as a good clock. But I tried to
>synchronize a juniper to another vendor's router that did answer with an=
 NTP
>response packet with a non-zero reference clock and it did synchronize.
>
>I presume it's a software bug in the server.

Luis,

Juniper can not act as a reference Server on it's own.
So you need to have a proper reference server to be
able to synchronize. This is how it is working since
day one and not a software bug.

I don't know if this was the case in the initial email.

regards
Josef

>  I'm using 5.1R2.4 worldwide.
>
>Unfortunately enough, one can't trace this with juniper's trace flags,=
 since
>there isn't any for NTP.
>
>Regards,
>
>-----------------------------------------------------------
>Luis Eduardo D=EDaz Zazo       Tel: +34 914 359 687
>Eurocomercial I&C S.A.       Fax: +34 914 313 240
>Valent=EDn Beato 5             mailto:lediaz@eurocomercial.es
>E-28037 Madrid
>Espa=F1a - Spain               http://www.eurocomercial.es
>-----------------------------------------------------------
>
>
>-----Mensaje original-----
>De: juniper-nsp-admin@puck.nether.net
>[mailto:juniper-nsp-admin@puck.nether.net]En nombre de Javier Alvira
>Enviado el: lunes 2 de septiembre de 2002 17:52
>Para: juniper-nsp@puck.nether.net
>Asunto: [j-nsp] NTP service
>
>
>Hi all
>I would like to know whether there is something special when configuring a
>NTP server in a Juniper (M-10, 5.0R1.4). This is what I configured:
>
>
>time-zone Europe/Lisbon;
>ntp
>
>     boot-server 10.128.0.20;
>     server 10.128.0.20;
>}
>-------------------------------------
>
>
>The "monitor traffic interface fe-0/0/0" shows this (10.128.0.20 is the NTP
>server, inside customer's network):
>15:25:48.737589 Out xx.yy.zzz.44.ntp > 10.128.0.20.ntp:  v3 client strat 0
>poll 6 prec -20 [tos 0x10]
>15:25:48.745207  In No-L2-hdr 10.128.0.20.ntp > xx.yy.zzz.44.ntp:  v3=
 server
>strat 1 poll 6 prec -19
>-----------------------------------
>
>
>But this is the output for "show ntp status":
>
>
>status=3Dc011 sync_alarm, sync_unspec, 1 event, event_restart,
>processor=3D"i386", system=3D"JUNOS5.0R1.4", leap=3D11, stratum=3D16,
>precision=3D-20, rootdelay=3D0.000, rootdispersion=3D6.105, peer=3D0,
>refid=3D0.0.0.0, reftime=3D00000000.00000000  Thu, Feb  7 2036 =
 6:28:16.000,
>poll=3D6, clock=3Dc119f9e3.d22c343b  Fri, Aug 30 2002 15:04:19.820,=
 state=3D1,
>phase=3D0.000, frequency=3D0.000, jitter=3D0.000, stability=3D0.000
>-----------------------------------------------------------
>
>
>So it seems it's not working... In fact, the initial difference of 16 secs
>(<128) has widened to 36 secs in 4 days...
>Any input will be welcome!
>Thanks,
>Javier
>
>_______________________________________________
>juniper-nsp mailing list juniper-nsp@puck.nether.net
>http://puck.nether.net/mailman/listinfo/juniper-nsp