On Aug 22, 2008, at 21:57, Udo van den Heuvel wrote:

> Ask Bjoern Hansen wrote:
>> If you perceive this to be in error (or fixed the server problem) and
>> want to add the server to the project again, please contact me or go
>> to http://www.pool.ntp.org/manage to use the automated admin
>> interface to cancel the deletion.
>
> You never proposed changes, fixes or even some process to detect a  
> cause.
> Meanwhile the monitoring system was observed by others to be  
> 'fragile'.
> So you automagically remove time sources.

Hi Udo (and everyone else),

I did spend some time on it when you wrote some time ago and I  
couldn't find any fault with the monitoring system related to this in  
the limited time I had to look at it.  Mostly I monitor the monitoring  
system by looking at the big picture.   There are only so many hours  
in a day and more than 1500 servers in the pool.   I look at 
http://www.pool.ntp.org/zone/ 
  and see that the proportion of "bad servers" have stayed relatively  
stable, even as we've been tightening the allowance for "bad time"  
over the years.

I do believe that I did encourage you to post here to the timekeepers  
list to get help; if I didn't I apologize.  Anyway, I'm glad it's  
brought up now - hopefully someone else with more time and insight  
will be able to track down what's going on.   My suggestion is to try  
tracking the time of your server from another server - both with a  
simple sntp client (ntpdate) and as a configured server in an ntpd  
setup.   With that setup there'll be more data to track the  
performance of your server and verify if it's the monitoring system  
being off or your system or network.

By the way - the pool system actually stores detailed logs for far  
longer than it shows by default.  You can add a limit= parameter

        http://www.pool.ntp.org/scores/80.101.128.228/log?limit=200

If I recall correctly the limit goes up to 5000 (please be gentle  
using that, it puts a bit of load on the system).

When I can't give more help it's basically because what's in that log  
is all the information I have -- and it's what the pool system already  
used to calculate the score - faulty or not.


  - ask

-- 
http://develooper.com/ - http://askask.com/


_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to