[time] Monitoring system

Guillermo Ballester Valor gbv
Wed Nov 12 19:17:38 UTC 2003


I'm the timekeeper of currently the worst server in the pool (as 
today at 6 UTC). :(

I've been trying to find the reason of such poor performance reported by the 
monitoring system. I changed the server, the router, and I couldn't find the 
reason. As you can see, it is declared unreachable most of the time.

On the other hand, I've been monitoring the traffic of NTP clients too. I use 
cron to analize ntp every minute. The script analizes the output of 'ntpdc 
-nc monlist' to select only non local clients activity and to detect 
suspicius gaps in the last reported activity queue. I can't see any gap. 
Clients are reaching the server continuosly. Just now it has about 60 clients 
with activity under 1024 secs. The logs of the script never detects gaps over 
30 secs in the range 0-128 secs. So, I can't explain why my server is lastly 
declared as unreachable.

As a aditional test, I usually check with 'ntpq -pn' a sever here in Spain 
sincronized with mine, and NEVER mine has declared as unreacheble, and reach 
is always '377'. (well, NEVER is not true, the server was really unreachable 
when I was changing hardware to fix the problem).

Any idea of what is the problem?.

Guillermo Ballester Valor
Linux user #117181. See http://counter.li.org/
gbv at oxixares.com
Ogijares, Granada  SPAIN

More information about the pool mailing list