[ntp:questions] Xntp d'ont slow the clock on alphaserver with Tru64

Tom Smith smith at cag.zko.hp.com
Fri Mar 14 13:43:59 UTC 2008


I see 2 possible problems:

 > daemon.log:Mar 14 07:58:27  xntpd[362529]: frequency initialized 0.000
 > from /var/ntp/drift

I would delete your driftfile after stopping (x)ntpd and
before restarting it. As it is, you are telling NTP to start with
a known drift rate of 0, and it will take a long(er) time to
compute an accurate rate. You are using a non-standard path
for that file on Tru64, so also make sure the ownership and
permissions are correct for the subdirectories you are using.
Each should be owned and writeable by root.

This may also be part of your problem:

 > 14 Mar 08:05:50 xntpd[362529]: get_kernel_info: ntp_adjtime() failed:
 > Function not implemented

I note that your version of ntp is a little old, so you might want
to update your patches, but I think the proximate cause of this
message is that your (x)ntpd image is trying to use the kernel time
discipline system call, but you have not enabled that in your kernel.

For the Tru64 kernel time discipline, /sys/conf/[SYSTEMNAME] should
contain the line:

   options NTP_TIME

If not, you can run "doconfig", making sure to include the NTP option
from the menu, or you can manually edit /sys/conf/[SYSTEMNAME]
to add the needed line and then run "doconfig -c [SYSTEMNAME]".
Both of these will build a new kernel containing the missing
routine.

-Tom

Stephane Guillou wrote:
> Hi,
> 
> I have configured an alphaserver with tru64 v5.1b to be NTP (ntpq
> 4.1.1 Fri Apr 29 14:25:45 EDT 2005) client, its configuration file is:
> # cat /etc/ntp.conf
> server 127.127.1.0
> fudge 127.127.1.0 stratum 10
> driftfile /var/ntp/drift
> multicastclient 224.0.1.1
> broadcastdelay 0.008
> authenticate no
> 
> This alphaserver synchronise its clock thanks to two stratum 2 NTP
> server connected on the same subnet.
> ntpq> peer
>      remote           refid      st t when poll reach   delay
> offset  jitter
> ==============================================================================
>  LOCAL(0)        LOCAL(0)        10 l   58   64  377    0.000
> 0.000   0.488
> +serv1        ntpclock-lan     2 -   32   64  377    0.488  -67.351
> 1.347
> *serv2         ntpclock-lan     2 -   16   64  377    0.488  -67.501
> 1.570
> 
> It seems that xntpd never slow the clock and a step is done as soon as
> the offset is under -128ms.
> daemon.log:Mar 14 07:58:27 xntpd[362529]: ntpd 4.1.1 Fri Apr 29
> 16:10:01 EDT 2005
> daemon.log:Mar 14 07:58:27  xntpd[362529]: precision = 976 usec
> daemon.log:Mar 14 07:58:27  xntpd[362529]: kernel time discipline
> status 0040
> daemon.log:Mar 14 07:58:27  xntpd[362529]: frequency initialized 0.000
> from /var/ntp/drift
> 
> The file /var/ntp/drift always contain 0.000. Reset and lost
> synchronisation are observed every 2 hours.
> 
> Using logfile option in /etc/ntp.conf I can also see this message
> 14 Mar 08:05:50 xntpd[362529]: get_kernel_info: ntp_adjtime() failed:
> Function not implemented
> 14 Mar 08:17:41 xntpd[362529]: time reset -0.149086 s
> 14 Mar 08:17:41 xntpd[362529]: system event 'event_clock_reset' (0x05)
> status 'leap_none, sync_unspec, 5 events, event_peer/strat_chg' (0x54)
> 14 Mar 08:17:41 xntpd[362529]: system event 'event_peer/
> strat_chg' (0x04) status 'leap_none, sync_unspec, 6 events,
> event_clock_reset' (0x65)
> 
> Is there someone to help me with this problem?
> 
> Regards
> 
> Steph.




More information about the questions mailing list