[ntp:questions] Re: tos orphan and poll frequencies

user at domain.invalid user at domain.invalid
Wed May 3 19:19:46 UTC 2006


Roger,

First, understand wht the orphan mode is for, configurations with no 
outside redundant source of time and where the survivors in a possibly 
damaged network can agree on which cow in the herd to follow. Note that 
each cow can have a small but significant local clock frequency error, 
so only one of them must be elected.

The tos orphan <stratum> command puts a ceiling <stratum> on the local 
stratum. If the stratum reaches that value the client switches the 
routing metric from stratum to a random value, presumably unique in the 
network. As long as connection paths exist in the network, even if 
broadast, the survivors will be configured to use the lowest metric and 
no loops will form. If one or more hosts are NTPing with bona fide 
Internet machines and the highest stratum is less than the orphan, the 
NTP subnet will operate as normal. In case all Internet sources are 
lost, the stratum ois set to the orphan stratum, and the network will 
reconfigure as an orphan.

Note this is not the expected behavior when tos stratum <stratum> is not 
used. In the ordinary case, if all sources are lost, the only thing that 
changes is the dispersion, which climbes to infinity at 16 PPM.

Dave

Lindholm Roger wrote:

> Hello,
> Anyone who knows if setting "tos orphan x" in any way changes poll frequencies (iburst, minpoll, maxpoll etc.) or does it only affect the stratum and selection algorithms? 
> If anyone has experience from using orphan mode on a grander scale I would be interested to hear about stability issues, some best practices (if any) what to do and not to do.
> Best regards
> // Roger
> _______________________________________________
> questions mailing list
> questions at lists.ntp.isc.org
> https://lists.ntp.isc.org/mailman/listinfo/questions
> 




More information about the questions mailing list