So I had a support request for an interesting case affecting 2 Cisco TelePresence endpoints both running TC 7.3.2 (the latest software at the time of writing this). One was an SX20, the other an older C20. Both of these units were configured to connect to the same NTP server in their local region and both were showing un-synchronized.
Initially I did what (i hope) most people would have done and blame the NTP server which was a Windows box and only these 2 endpoints were configured to use it.
Note: Windows NTP boxes are considered not supported by Cisco because by default they do not provide a full NTP service and only offer a limited version, this is sometimes referred to as Simple NTP or SNTP. Windows machines can be configured to run full NTP using a few registry changes as outlined in a great Cisco support forum entry here. (While this should work it will still not be supported by Cisco TAC).
Even after making these changes the issue persisted but we tested a 3rd party NTP client and this connected to the NTP server correctly from the same subnet.
I connected to both VC units via CLI and was shocked to see that the ‘xconfig’ output showed the NTP server configured as a null value even though it was configured correctly on the GUI.
Setting the values on the CLI manually also did not fix the problem – rebooting the VC units and re-entering the NTP server details on the GUI resolved the problem (the changes were reflected in the CLI).
Two lessons I took away from this:
- Use xconfig and xstatus commands via the CLI to verify configuration on VC endpoints – don’t rely on the GUI
- If in doubt reboot (I wish this was the first time that I learned this but I seem to need to keep reminding myself on the truth behind this statement!)