NTP error

[gb-fairchild]

I'm getting this from RTMT:

At Mon Jan 31 20:30:25 EST 2011 on node 10.164.0.11, the following SyslogSeverityMatchFound events generated: SeverityMatch - Critical ntpRunningStatus.sh: The local NTP client is off by more than the acceptable threshold of 3 seconds from its remote NTP system peer. The normal remedy is for NTP Watch Dog to automatically restart NTP. However, an unusual number of automatic NTP restarts have already occurred on this node. No additional automatic NTP restarts will be done until NTP time synchronization stabilizes. Check NTP status via CLI 'utils ntp status' and ensure good network connectivity to the remote NTP server(s). After rectifying the problem, restart NTP on this node and any dependent nodes via CLI 'utils ntp restart'. If all else fails, schedule a reboot of this server.

When I check status from CLI on publisher  (CUCM 7.1.3) I get:

admin:utils ntp status
ntpd (pid 20391) is running...

     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
*127.127.1.0     LOCAL(0)        10 l    5   64  377    0.000    0.000   0.004
x10.0.17.3       10.0.17.4        3 u  127 1024  377    0.572  -119932 229.778
x10.0.96.93      .GPS.            1 u  185 1024  377    7.986   -1.865   1.466

synchronised to local net at stratum 11
   time correct to within 11 ms
   polling server every 1024 s

Current time in UTC is : Tue Feb  1 01:34:46 UTC 2011
Current time in America/New_York is : Mon Jan 31 20:34:46 EST 2011

Not sure what this is telling me - but it says time is correct to within 11ms

Thanks,

Bill

[htluo · Accepted solution]

It cannot synchronize to 10.0.17.3 because the offset was too huge. It's using the local clock right now.

Try restart the NTP service from CLI: utils ntp restart

Michael

http://htluo.blogspot.com

[srsivara]

In addition to Michael's suggestion, can you let us know if 10.0.17.3 a Windows NTP server or IOS-based ? Please keep in mind the following documentation for Linux based CallManagers :

CSCte17541    SRND to specifically mention that Windows NTP server is not supported

- Sriram

Please rate helpful posts !

[htluo]

I think Cisco abuses the word "not supported".

In Cisco's dictionary, "not supported" means "I'm not going to help you on this".    It doesn't mean "It will not work".

Actually Windows NTP works with CUCM (and other UC appliances) perfectly.  The caveat was: Windows has a couple registry keys as NTP parameters.  One of them is called "LocalClockDispersion", which is an indicator of clock accuracy.  Windows was modest enough to set this to 1000 (ms) by default, which is not acceptable from UC appliance point of view.  All you need to do is to set it to 0 in registry.

For more details, see my blog at http://htluo.blogspot.com/2009/02/ntp-network-time-protocol.html

Thanks!

Michael

[gb-fairchild]

I agree - I was just thinking that this has been running for 5 years using the Windows NTP without any problems. It seems to be happy again after the restart.  I will check into that registry setting.

Thanks for your help.

Bill

[srsivara]

+5 for that explanation on windows NTP servers, Michael !

- Sriram