cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
3700
Views
12
Helpful
9
Replies

NTP gets time but won't update - status/debug shows ref clock 0.0.0.0

schm196
Level 1
Level 1

My NTP config on a 4908G-L3 switch router looks like this:

ntp source Port-channel1

ntp update-calendar

ntp server 10.1.2.35 prefer

ntp server 10.1.2.1

Here is the current situation:

NM-Core#sh ntp asso

address ref clock st when poll reach delay offset disp

~10.1.2.35 0.0.0.0 15 336 1024 377 1.7 160991 0.6

~10.1.2.1 0.0.0.0 15 305 1024 377 1.6 160948 1.8

* master (synced), # master (unsynced), + selected, - candidate, ~ configured

NM-Core#sh ntp stat

Clock is unsynchronized, stratum 16, no reference clock

nominal freq is 250.0000 Hz, actual freq is 250.0000 Hz, precision is 2**24

reference time is 00000000.00000000 (16:00:00.000 PST Wed Dec 31 1899)

clock offset is 0.0000 msec, root delay is 0.00 msec

root dispersion is 0.00 msec, peer dispersion is 0.00 msec

Here's the packet debugging output:

.Oct 15 17:01:45.274: NTP: xmit packet to 10.1.2.35:

.Oct 15 17:01:45.274: leap 3, mode 3, version 3, stratum 0, ppoll 1024

.Oct 15 17:01:45.274: rtdel 0000 (0.000), rtdsp 10001 (1000.015), refid 00000000 (0.0.0.0)

.Oct 15 17:01:45.274: ref 00000000.00000000 (16:00:00.000 PST Wed Dec 31 1899)

.Oct 15 17:01:45.274: org C156C919.42D0E513 (10:04:25.260 PDT Tue Oct 15 2002)

.Oct 15 17:01:45.274: rec C156C878.46CE4270 (10:01:44.276 PDT Tue Oct 15 2002)

.Oct 15 17:01:45.274: xmt C156C879.466846A4 (10:01:45.275 PDT Tue Oct 15 2002)

.Oct 15 17:01:45.274: NTP: rcv packet from 10.1.2.35:

.Oct 15 17:01:45.274: leap 0, mode 4, version 3, stratum 15, ppoll 1024

.Oct 15 17:01:45.274: rtdel 0000 (0.000), rtdsp 0000 (0.000), refid 00000000 (0.0.0.0)

.Oct 15 17:01:45.274: ref C156C908.3AE1476A (10:04:08.229 PDT Tue Oct 15 2002)

.Oct 15 17:01:45.274: org C156C879.466846A4 (10:01:45.275 PDT Tue Oct 15 2002)

.Oct 15 17:01:45.274: rec C156C91A.42D0E513 (10:04:26.260 PDT Tue Oct 15 2002)

.Oct 15 17:01:45.274: xmt C156C91A.42D0E513 (10:04:26.260 PDT Tue Oct 15 2002)

.Oct 15 17:01:45.274: inp C156C879.46D23BFB (10:01:45.276 PDT Tue Oct 15 2002)

This happens 16 times in a row - you can see that the switch transmits to the correct IP address, receives the corrected time (about 3 minutes off) but then doesn't update it's own clock, probably because "reference clock 0.0.0.0" I assume. My NTP servers are Windows 2000 machines running NetTime 2.0b6 synchronized to 5 outside stratum 1 sources. any ideas why the switch refuses to update itself or how to get the "reference clock 0.0.0.0" changed? Thanks!

9 Replies 9

schm196
Level 1
Level 1

NTP version protocol used on W2K servers (NetTime 2.0b6):

SNTP port 123

This applies to how the server itself communicates with the external stratum 1 servers; I do not know whether this affects how aother machines can synchronize with the servers. NetTime itself is supposedly fully compliant with NTP standards through version 3.

Other available options include:

RFC868 (TCP) port 37

RFC868 (UDP) port 37

I don't know if these will work for synchronizing my servers to the outside world; haven't tried them yet.

ggersch
Level 1
Level 1

The problem is not in your router config. The NetTime software is not providing the correct ntp info to the routers. If you look at the output of your 'sho ntp assoc' command, you will see that they are outputting time with a stratum of 15. That puts your router at stratum 16, which is unsynchronized. If the NT servers are syncing to stratum 1 servers, they should be at stratum 2, and your routers would be stratum 3.

Also, check the time on both your router and the servers. That reference time year of 1899 is highly suspect as well.

Unless the NetTime software can be configured to output stratum 2 signals, it won't work. The port 37 unix time standard is not supported as an ntp source. The native Windows w32time will not act as a server, so don't use it either.

I've had excellent experience with the real xntpd daemon port for NT servers. You can get the latest NT versions at: http://www.five-ten-sg.com/ Look for the 'NTP 4.1.71' link bottom left.

Greg

Thanks a lot; I think that's the problem (not to mention a bug in NetTime). NetTime has a checkbox to "allow other computers to synch to this machine", which -if left unchecked- I believed would set the response to stratum 15 with the same result as described by you. I have them checked on my servers, though, so something must be crooked with the response. It is interesting, however, that you pointed out the 1899 reference transmission by the router in the first place. Where could this possibly come from? I had previously set the router's clock manually to a fairly accurate time, including the year.

Anyway, thanks again for the helpful response. I will try the xntpd daemon port for NT, which hopefully works just as well on 2K and XP.

- Matthias.

About xntpd... Looks like it can do everything an engineer and command-line fanatic could ever want to accomplish but I really didn't set out to become a Ph.D.-level expert of NTP. The reason we chose NetTime previously was that it is extremely easy to set up, configure, and monitor - I feel right now like I would have to devote several days of full-time studying into getting something up and running as desired using xntpd. The simple install works but everything after that (monitoring, configuration changes) is a mystery. Do you know of any other applications that might fit my needs (and level of expertise) better? Thanks.

The xntpd software is not as daunting as it looks. It installs easily and runs as a service. The only real config you have to do once its installed is to add your servers to the ntp.conf file. Use notepad to edit the file. If it doesn't sync to one of your servers, you might need to add a 3 to specify a v3 server since this is the v4 daemon.

To monitor it, use the ntptrace.exe and ntpq.exe commands. You can copy these to your workstation. They are command line, but very easy. 'ntptrace' has one command option, which is the name of the server if its not being run on the server.

ntptrace servername

'ntpq' lets you query the ntp server interactively. Its very similar to the ntp commands in IOS. There are lots of commands, but the only ones you need are: 'host' to set the ntp server to talk to, 'assoc' to see the status, and 'peers' to see the statistics. It feels a lot like nslookup if you know that tool.

If you can do IOS, you can do the xntpd.

Greg

Thanks for the encouragment and, yes, it wasn't that hard after all - it just looked very low-level at the beginning. I have it up and running on my own gateway servers now; it works great. The Cisco devices synchronize without a hitch and for my other Windows devices I can still use NetTime. Thanks again for the help! - Matthias.

Hi Greg

How we conf the switch or the router to ntp in terms of command-line-interface

Can you give me an example?

Thanks in Advance

MP

My NTP config on a 4908G-L3 switch router looks like this:

ntp source Port-channel1

ntp update-calendar

ntp server 10.1.2.35 prefer

ntp server 10.1.2.1

>>> Of course, replace the server ip addresses with your own preferred sources. As a matter of good practice, do not have all of your network's nodes hit an external (free) stratum 2 server, just one or two - then synchronize all of your own systems to these now internal sources. <<<

- Matthias

jeff_caprock
Level 1
Level 1

Bypass the W2K server and try these two servers instead, all are stratum 2 servers and work for our network:

192.5.41.209, 192.5.41.40, and 132.163.4.102

Our perimeter router is directed to these three, and all internal devices take time updates from this perimeter router instead of a Windows server. We've seen this issue before with some ntp incarnations that use SNTP instead of NTP.

-Jeff

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: