[ntp:questions] Re: Is ntp really what I need?

Richard B. Gilbert rgilbert88 at comcast.net
Mon Feb 14 16:48:05 UTC 2005


Paul Hilton wrote:

>Hello:
>
>I am using SuSE 9.0, I start ntpd on boot, and I have problems with my
>clock:
>
>I do not need a very accurate clock, but it would be nice to keep my
>network within a few seconds at worst, for file synchronization.
>
>The system clock seems to be based upon an interrupt driven counting
>method, there is a CMOS clock in the hardware that seems to work very
>well, but it is largely ignored except for boot up.
>
>When I run a job that strains the limits of the processor (such as playing
>some videos) the clock accumulates a drastic negative error, I presume
>that the clock interrupts are being missed. This makes a mockery of the
>concept of drift.
>
>Sometimes ntpd corrects this, but quite often ntpd just dies (I presume
>because of the repeated 'insane' errors). The recent log shows:
>
>11 Feb 12:42:12 ntpd[1792]: time reset 6.183748 s
>11 Feb 12:42:12 ntpd[1792]: synchronisation lost
>11 Feb 13:08:08 ntpd[1792]: time reset 3.694851 s
>11 Feb 13:08:08 ntpd[1792]: synchronisation lost
>11 Feb 14:39:59 ntpd[1792]: time reset 84.642767 s
>11 Feb 14:39:59 ntpd[1792]: synchronisation lost
>11 Feb 15:01:43 ntpd[1792]: time reset 9.474174 s
>11 Feb 15:01:43 ntpd[1792]: synchronisation lost
>12 Feb 01:07:49 ntpd[1792]: time correction of 1008 seconds exceeds sanity
> limit (1000); set clock manually to the correct UTC time.
>14 Feb 10:33:12 ntpd[21757]: signal_no_reset: signal 17 had flags 4000000
>14 Feb 10:33:12 ntpd[21755]: running as uid(74)/gid(65534) euid(74)/egid(65534).
>14 Feb 10:33:12 ntpd[21755]: frequency initialized 86.927 from 
> /var/lib/ntp/drift/ntp.drift
>14 Feb 10:36:35 ntpd[21755]: kernel time discipline status change 41
>14 Feb 10:53:42 ntpd[21755]: time reset 0.656745 s
>14 Feb 10:53:42 ntpd[21755]: synchronisation lost
>
>
>Would I just be better off running ntpd -q hourly as a cron job?
>
>My ntp.conf is:
>
>server 127.127.1.0
>fudge 127.127.1.0 stratum 10
>server blade.avnf.com
>driftfile /var/lib/ntp/drift/ntp.drift
>logfile /var/log/ntp
>peer 192.168.0.6
>
>Thanks for any insights,
>Paul Hilton
>somedomain might be yahoo
>  
>
I'm afraid you are right about your system losing clock interrupts.  
Unfortunately ntpd must rely on the proper operation of the clock.  It 
can correct minor errors but it can do nothing in the face of an O/S 
that gives higher priority to playing videos than to updating the 
clock.  Change to a better O/S or play your videos somewhere else.



More information about the questions mailing list