Large NTP offset, clock off by 4 minutes

[mralston]

I've set up NTPD on a newly installed system running CentOS 5.7.

My ntp.conf file is as per the default that CentOS ships with and other than experimenting with a few different NTP servers I haven't touched it. I'm currently using the default CentOS NTP servers.

I've found that the clock is consistently 4 minutes fast. ntpq -p shows a very large offset against each server...

# ntpq -p
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
+mighty.poclabs. 72.14.177.132    3 u   31   64  377    3.904  -240033  24.871
+64.73.32.134    18.26.4.105      2 u   44   64  377   20.950  -240034  20.516
+name3.glorb.com 216.218.254.202  2 u   26   64  377   52.829  -240008  16.164
 LOCAL(0)        .LOCL.          10 l   13   64  377    0.000    0.000   0.001

I presume this offset is the cause of the problem... 240033 milliseconds is 4 minutes.

Any ideas how to fix this?

Thanks,

- Matt

[Irithori]

You can try to shutdown ntpd,
then

ntpdate -b <NTP_SERVER_OF_YOUR_CHOICE>

and start ntpd again.

This should set the time right and ntp will make sure it stays that way.

If not, is that a VM by any chance?

[mralston]

Hi Irithori.

Thanks for the info. Unfortunately ntpdate -b didn't help.

# ntpdate -b ntp.your.org
 5 Oct 07:01:54 ntpdate[12237]: step time server 204.9.54.119 offset -243.928245 sec

The server is a VM. Running on Rackspace Cloud under Xen.

I did consider it might be a VM issue, but their support department don't seem to think there should be a problem, and my other VMs don't have this issue... and yet this one does. Go figure!

- Matt