[ntp:questions] NTP in a Linux cluster

David J Taylor david-taylor at blueyonder.not-this-part.nor-this.co.uk.invalid
Tue Sep 8 06:18:40 UTC 2009


"Lorcan" <> wrote in message 
news:bffea289-aa6a-4b2a-8bac-9a6bfe93fcde at y36g2000yqh.googlegroups.com...
> Folks,
>
> Could any NTP experts suggest how I should best configure NTP in
> a loosely-coupled Linux cluster, where intra-cluster synchronization
> is
> the top priority?
[]
> The key requirement here is the as-close-as-possible synchronization
> between
> nodes in the cluster; that is far more important than closely tracking
> the external
> NTP server(s).
[]
> It would be preferable if the configuration of all nodes could be
> identical.   A solution
> that requires two or more node types ("master" and "slave" perhaps)
> with different
> settings would be acceptable, of course, if the preferred one-size-
> fits-all approach
> is impossible.
[]
> Any and all answers would be gratefully received!
>
> Thanks in advance,
>
> Lorcan Hamill

Speaking generally, not about Linux clusters, I have been pleased with 
using GPS to sync PCs.  If you can get multiple GPS receivers connected to 
nodes in the cluster and have them peer with each other, that could be a 
good solution, without a single point of failure.  I have had some success 
feeding multiple PCs from a single GPS receiver, and you could so that as 
well using some sort of RS232 booster.

When you say: "as-close-as-possible synchronization" - what does that mean 
for you.  Seconds?  Milliseconds?  Microseconds?

According to:
  http://doc.ntp.org/4.2.0/ntpd.html

  "If the -x option is included on the command line, the clock will never 
be stepped and only slew corrections will be used."

Cheers,
David 




More information about the questions mailing list