[ntp:questions] Using two NTP Server: Bad?

unruh unruh at wormhole.physics.ubc.ca
Thu Apr 21 02:57:26 UTC 2011


On 2011-04-21, Ben Rockwood <benr at cuddletech.com> wrote:
> I've read in the past that clients should always have an odd-number of
> NTP servers; 1 server or 3 servers but not 2.  If I recall the reason
> was that clients could become "confused" and needs a tie breaker.
>
> Question 1: I want to challenge this old assumption.  Is there truth to
> it? 
>
> In many environments there is a desire to have 2 NTP servers for
> redundancy purposes, but not 3 due to limited resources.

If one of the servers disappears, the other can take over. If one goes
nuts and begins to deliver the wrong time, your system has no idea which
of the two is right. 3 allows 2 to outvote one. 

>
> Question 2: Furthermore, if you have 2 local NTP servers is it
> preferable to have them sync off of different sources to avoid a client
> syncing servers that are using the same reference clock?  ie: Is this bad?:
>
> $ ntpq -pn
>      remote           refid      st t when poll reach   delay   offset 
> jitter
>==============================================================================
> *10.0.91.10      207.171.7.152    3 u  756 1024  377    0.494    0.596  
> 0.352
> +10.0.91.82      207.171.7.152    3 u  953 1024  377    0.135    0.144  
> 0.241

Well, it depends on whether or not 207.171.7.152 begins to deliver the
wrong time. If it does both of yours do as well.


>
>
>
> Thank you.
>
> benr.




More information about the questions mailing list