[ntp:questions] Peering recommendation

Richard B. Gilbert rgilbert88 at comcast.net
Fri Jun 29 14:13:14 UTC 2007


ulf.norberg at banverket.se wrote:
> Hi.
> 
> I've been searching the net for recommendation about peering and how to
> avoid synchronisation loops, but I haven't found anything that answers
> my questions.
> 
> Almost every time I see a picture about NTP, peering is drawn like my
> "case1" below, but I never seen an explenation saying, it must look like
> this or you have to do it in a certain way.
> 
> Can peering be done as described in my case2 and case3 or should anyone
> of them be avoided?
> 
> Case1
> |-------|       |-------|       |-------|       |-------|
> |server1| <---> |server2| <---> |server3| <---> |server4|
> |-------|       |-------|       |-------|       |-------|
>     
> 
> 
> Case2
> |-------|       |-------| 
> |server1| <---> |server2| 
> |-------|       |-------| 
>     ^               ^
>     |               |
>     |               |
>     v               v
> |-------|       |-------|
> |server3| <---> |server4|
> |-------|       |-------|
> 
> 
> Case3
> |-------|       |-------| 
> |server1| <---> |server2| 
> |-------|<     >|-------| 
>     ^     \   /     ^
>     |      \ /      |
>     |       X       |
>     |      / \      |
>     v     /   \     v
> |-------|<     >|-------|
> |server3| <---> |server4|
> |-------|       |-------|
> 
> Best regards
> Uffe
> _______________________________________________
> questions mailing list
> questions at lists.ntp.isc.org
> https://lists.ntp.isc.org/mailman/listinfo/questions
> 

Forget the diagrams and concentrate on the configuration files!

For example, on server1:

peer server2
peer server3
peer server4

On server2
peer server1
peer server3
peer server4

etc.  Draw your own picture if you must, I won't even attempt it!

FWIW and IMHO, to whatever exent my opinions might be "humble" :-), each
peered server needs at least one unique source of time.  Without that 
unique source of time the peers can all be wrong in the same way and 
have no means of detecting the problem.




More information about the questions mailing list