ntp issue.

Herta Van den Eynde herta.vandeneynde at gmail.com
Thu Aug 2 22:28:30 UTC 2007


On 02/08/07, Andrew.Bridgeman at corusgroup.com
<Andrew.Bridgeman at corusgroup.com> wrote:
>
>
>
> Thanks for the info. I have done the things you mentioned but it still
> seems to be losing time. For example I have been doing the testing on one
> machine, I have made the changes you said, stopped started the ntpd service
> and within a couple of hours it was 1.6 sec behind the Windows servers it
> is time slaving from. My guess would be that if i leave it for a week it
> will be a least couple of hours behind.
> I have checked the drift file contents and it is just set to 0.000 and also
> the key file does not have anything in it apart from the default values. It
> seems to have a problem after the service have been restarted ( maybe issue
> polling and retrieving the information ) because when this is done it
> communicates with the server and the time it spot on.
>
> Can anyone suggest something else to try!
>
> Regards
>
> [IMAGE]
> |+----------------------------+-------------------------------------------|
> ||   "Herta Van den Eynde"    |                                           |
> ||   <herta.vandeneynde at gmail.|   To: "General Red Hat     |
> ||   com>                     |   Linux discussion list"                  |
> ||   Sent by:                 |   <redhat-list at redhat.com>                |
> ||   redhat-list-bounces at redha|   cc:                             |
> ||   t.com                    |   Subject: Re: ntp issue.  |
> ||                            |                                           |
> ||   31/07/2007 11:37         |                                           |
> ||   Please respond to General|                                           |
> ||   Red Hat Linux discussion |                                           |
> ||   list                     |                                           |
> ||                            |                                           |
> |+----------------------------+-------------------------------------------|
>
>
>
>
>
>
> On 31/07/07, Andrew.Bridgeman at corusgroup.com
> <Andrew.Bridgeman at corusgroup.com> wrote:
> >
> >
> > The ntp information is below. The Full IP address is show below just as
> > "IP" for security reasons.
> >
> > restrict default ignore
> > restrict IP mask 255.255.255.255 nomodify notrap noquery
> > restrict 127.0.0.1
> > server IP
> > fudge 127.127.1.0 stratum 10
> > driftfile /var/lib/ntp/drift
> > broadcastdelay 0.008
> > authenticate yes
> > keys /etc/ntp/keys
> >
> > Let me know what you think?
> >
> > Regards
> >
> > Andrew Bridgeman
> >
> >
> |+----------------------------+-------------------------------------------|
> > || "Herta Van den Eynde" |
> |
> > || <herta.vandeneynde at gmail.| To: "General Red Hat |
> > || com> | Linux discussion
> list" |
> > || Sent by: |
> <redhat-list at redhat.com> |
> > || redhat-list-bounces at redha| cc: |
> > || t.com | Subject: Re: ntp issue. |
> > || |
> |
> > || 31/07/2007 11:06 |
> |
> > || Please respond to General|
> |
> > || Red Hat Linux discussion |
> |
> > || list |
> |
> > || |
> |
> >
> |+----------------------------+-------------------------------------------|
> >
> >
> >
> >
> >
> >
> > On 31/07/07, Andrew.Bridgeman at corusgroup.com
> > <Andrew.Bridgeman at corusgroup.com> wrote:
> > >
> > >
> > >
> > >
> > > Hello All,
> > > I have a strange issue at the moment with regards ntp
> > > working incorrectly on 35 of our Redhat WS Version 3 Linux machines. I
> > have
> > > set them all up the same way so they use the IP of our Windows Cluster
> > > machine I did this by adding the IP to the ntpservers file and the
> > > tick-servers file. In addition edited the ntp.conf file to show the
> > cluster
> > > IP as well.
> > >
> > > What I seem to get is time differences on the machines within a day of
> > > stopping and starting the ntpd service. For example I did this
> yesterday
> > > and below shows all the machine times now ( 24 hours ) later, you will
> > > notice there are 30 sec to a minute differences already and I bet with
> in
> > > a week or a month they get even more out of sync. Has any body got any
> > > ideas on what i can check or change to get this working correctly as
> > > currently it is causing real issues when we run analysis jobs over
> night.
> > >
> > > Time on Machines
> > > ==============
> > > Tue Jul 31 10:38:46 BST 2007
> > > Tue Jul 31 10:38:48 BST 2007
> > > Tue Jul 31 10:38:47 BST 2007
> > > Tue Jul 31 10:38:48 BST 2007
> > > Tue Jul 31 10:38:48 BST 2007
> > > Tue Jul 31 10:39:00 BST 2007
> > > Tue Jul 31 10:39:04 BST 2007
> > > Tue Jul 31 10:39:04 BST 2007
> > > Tue Jul 31 10:39:04 BST 2007
> > > Tue Jul 31 10:39:05 BST 2007
> > > Tue Jul 31 10:39:06 BST 2007
> > > Tue Jul 31 10:39:06 BST 2007
> > > Tue Jul 31 10:38:58 BST 2007
> > > Tue Jul 31 10:38:59 BST 2007
> > > Tue Jul 31 10:38:59 BST 2007
> > > Tue Jul 31 10:39:00 BST 2007
> > > Tue Jul 31 10:39:00 BST 2007
> > > Tue Jul 31 10:39:01 BST 2007
> > > Tue Jul 31 10:39:01 BST 2007
> > > Tue Jul 31 10:39:02 BST 2007
> > > Tue Jul 31 10:39:02 BST 2007
> > > Tue Jul 31 10:39:02 BST 2007
> > > Tue Jul 31 10:39:02 BST 2007
> > > Tue Jul 31 10:39:05 BST 2007
> > > Tue Jul 31 10:39:04 BST 2007
> > > Tue Jul 31 10:39:04 BST 2007
> > > Tue Jul 31 10:39:04 BST 2007
> > > Tue Jul 31 10:39:05 BST 2007
> > > Tue Jul 31 10:39:05 BST 2007
> > > Tue Jul 31 10:39:05 BST 2007
> > > Tue Jul 31 10:39:07 BST 2007
> > > Tue Jul 31 10:39:07 BST 2007
> > > Tue Jul 31 10:39:07 BST 2007
> > > Tue Jul 31 10:39:08 BST 2007
> > > Tue Jul 31 10:39:12 BST 2007
> > >
> > > Regards
> > >
> > > Andrew Bridgeman
> >
> > What does your ntp config look like? (grep -v ^\# /etc/ntp.conf)
> >
> > Anything in the ntp log?
> >
> > Kind regards,
> >
> > Herta
> >
>
> Try putting your restrict clauses towards the bottom of the file.
> Also verify that the drift file exists, and that it's owner and group
> are set to ntp. Also check syslog for ntp errors.
> FWIIW, I usually find it easier to write ntp logs to an alternate log
> file, which you can do by adding a clause such as this one:
>
> logfile /var/log/ntp # alternate log file
>
> Kind regards,
>
> Herta
>
Hi Andrew,

You write that after restarting ntpd, "within a couple of hours it was
1.6 sec behind" and a bit further: "It seems to have a problem after
the service have been restarted ( maybe issue polling and retrieving
the information ) because when this is done it
communicates with the server and the time it spot on."
I'm confused.  It sounds like at a certain point ntpd works properly.
Or what do you mean by "when this is done"?

Also, could you let us know whether there's anything in either syslog
or in an ntp log file?

Kind regards,

Herta




More information about the redhat-list mailing list