Subject: Re: ntp is very slow to be operational
To: Thierry Lacoste <th.lacoste@wanadoo.fr>
From: rudolf <netbsd@eq.cz>
List: netbsd-help
Date: 10/25/2005 14:00:49
Thierry Lacoste wrote:
> Sorry, I didn't wait enough:
> Oct 25 12:29:47 vega ntpd[277]: frequency error 512 PPM exceeds tolerance 500 
> PPM
> 
> Is it safe to use this machine as the time server in a production environment?
> I use kerberos so I need the machines to be in sync with one another;
> not necessarily with the outer world.
> 
> Regards,
> Thierry.

Hi,

I've one dual-opteron machine in a 'soft production environment' :) as a 
ntp *client* and so far I can tell that it is not panicking, if this is 
what you want to hear. It is synchronized to one local ntp server, the 
message you mentioned I can see aproximately once per hour, system is 
NetBSD 2.1_RC6. The current uptime is not very big, but you can get the 
picture:

$ uptime
  1:53PM  up 5 days,  5:21, 1 user, load averages: 8.58, 8.59, 8.67
$ ntpq -p
      remote           refid      st t when poll reach   delay   offset 
  jitter
==============================================================================
*our.local.ntp.server 195.113.144.201  2 u   89  128  377    0.135 
-17.857  14.012
$

I have been told that the message 'could' indicate a hw problem.

r.