Question About NTP Synchronization

Latest response

Hi all,
I have a question about NTP synchronization. If we have multiple NTP servers in our internal network (3 or more) and first server is synchronizing with an external NTP servers, GPS or other reference time. Second server is synchronizing with first server and third server with second server, then what happened if each server can not communicate with upper layer?
Clients are synchronizing with third server and if third server couldn't communicate with upper layer, what will be happened for clients?
This is not real scenario but I want to understanding impact of internet outage or any other connectivity outage on our clients.
Linux clients showing * left of NTP server address and it's an indicator about NTP server communication.
What does it mean, if * is not showing but NTP server was reachable on network?

Responses

Each NTP server has a "stratum" value where a GPS/atomic clock is Stratum 0, and NTP servers grow their Stratum number by 1 each "hop" they are away from a Stratum 0 source.

Off the top of my head, I bet if you had a set of three cascading NTP servers where one is Stratum 1, one is Stratum 2, and one is Stratum 3, clients would probably always prefer the Stratum 1 server and ignore the rest. So I think the setup you describe probably isn't worth the bother, you may as well just setup one NTP server.

Ideally you'd have multiple servers in the same Stratum so clients can get an accurate source of the real time. Say one server becomes unavailable then the other servers can be used.

If you are concerned about the upstream clock becoming unavailable, then you could configure your NTP servers to getting their upstream clock from a different source or an NTP pool.

Thank you Jamie.