[Beowulf] ntpd wonky?

Nifty Tom Mitchell niftyompi at niftyegg.com
Wed Dec 17 17:05:36 PST 2008

On Sun, Dec 14, 2008 at 04:37:33PM -0800, Greg Lindahl wrote:
> On Thu, Dec 11, 2008 at 11:01:31PM -0800, Bernard Li wrote:
> > Have you tried other pools eg. pool.ntp.org?
> That is stable for me. So it's not me, it's Red Hat's pool that's
> wonky.
> I see that CentOS switched to using ntp.org in 5.2, which I didn't
> automagically get thanks to rpm creating ntpservers.rpmnew, even
> though I hadn't modified the ntpservers file. Mmf.

Does it make sense to have a small local pool between the
local cluster and the internet?

A couple of venerable 'clock[1,2,...]' boxes with a single network
interface sitting on a DMZ could have a buffering effect for a thousand
boxes one NTP stratum behind them.  Major co-location sites should have 
some provisioning for quality local NTP time references as well.

Given the design of pool.ntp.org it makes sense for a business to 
do some quality audits of ntp.org hosts and contact local high
quality ones directly.

	T o m  M i t c h e l l 
	Found me a new hat, now what?

More information about the Beowulf mailing list