[time] Help diagnose 72.64.118.118

Jeffrey Goldberg jeffrey
Thu Apr 24 18:39:08 UTC 2008


I was surprised to get a "your server is below the threshold" notice  
this morning, and a quick investigation turns nothing up that I can  
see wrong at my end (other than some log messages that might suggest  
frequent restarts of ntpd.  See below.)

However, the graphs at

  http://www.pool.ntp.org/scores/72.64.118.118#graph_explanation

aren't pretty with respect to my offset.  If I understand the graphs  
correctly, my system has been perfectly reachable, but has been having  
fits of providing poor quality time.

So I like help understanding what I need to fix.

Here is some information about my system:

$ ntpq -c 'rv 0'
status=06f4 leap_none, sync_ntp, 15 events, event_peer/strat_chg,
version="ntpd 4.2.0-a Sun Mar  9 12:48:14 CDT 2008 (1)",
processor="i386", system="FreeBSD/7.0-RELEASE", leap=00, stratum=3,
precision=-18, rootdelay=219.881, rootdispersion=1338.284, peer=54487,
refid=17.254.0.31,
reftime=cbbad3b8.cda7a1a2  Thu, Apr 24 2008  4:39:04.803, poll=15,
clock=cbbb4c82.615ab5ae  Thu, Apr 24 2008 13:14:26.380, state=4,
offset=2.108, frequency=-31.515, jitter=11.246, stability=0.322

And here is my /etc/ntp.conf

=============================
# /etc/ntp.conf for dobby
# $Id: ntp.conf,v 1.2 2007/02/26 18:39:21 root Exp root $

# set up drift file
# driftfile /var/db/ntp.drift # default on FreeBSD

# set up rate limiting
discard
restrict default limited kod

## Undisciplined Local Clock. This is a fake driver intended for backup
## and when no outside source of synchronized time is available.
##
server 127.127.1.0              # local clock (LCL)
fudge  127.127.1.0 stratum 10   # LCL is unsynchronized

## Outside source of synchronized time
##
## server xx.xx.xx.xx           # IP address of server
#server pool.ntp.org
server ntppub.tamu.edu minpoll 5 maxpoll 12
server ntp.dayww.net minpoll 5 maxpoll 12
server time.apple.com minpoll 5 maxpoll 17
server ntp2.kansas.net minpoll 5 maxpoll 14
server time.utdallas.edu minpoll 5 maxpoll 12


## internal sources
# gecko
peer 10.1.10.131 minpoll 4 maxpoll 9
# hagrid
peer 10.1.10.136 minpoll 4 maxpoll 9
# ron
peer 10.1.10.133 minpoll 4 maxpoll 9
# harry
peer 10.1.10.132 minpoll 4 maxpoll 9
=============================

Please note that some of the servers I use in that file are "by  
permission" and not necessarily mean as fully public stratum 2  
servers.  So please be respectful of that information.

The system has been up for 45 days.

And here is an excerpt of grep ntpd log-file  (log times are US CDT,  
which I think is UTC-6:00.  I swear the next system I build will run  
in UTC for easier log coordination).


Apr  1 13:54:50 dobby ntpd[593]: kernel time sync enabled 2009
Apr  1 15:03:06 dobby ntpd[593]: kernel time sync enabled 6009
Apr  4 12:27:58 dobby ntpd[593]: kernel time sync enabled 2009
Apr  4 14:09:30 dobby ntpd[593]: kernel time sync enabled 6009
Apr  5 12:21:47 dobby ntpd[593]: kernel time sync enabled 2009
Apr  5 14:03:05 dobby ntpd[593]: kernel time sync enabled 6009
Apr 12 08:13:25 dobby ntpd[593]: kernel time sync enabled 2009
Apr 12 09:00:01 dobby ntpd[593]: too many recvbufs allocated (40)
Apr 12 09:21:41 dobby ntpd[593]: kernel time sync enabled 6009
Apr 15 22:42:03 dobby ntpd[593]: kernel time sync enabled 2009
Apr 15 23:14:28 dobby ntpd[593]: kernel time sync enabled 6009
Apr 17 14:31:53 dobby ntpd[593]: kernel time sync enabled 2009
Apr 17 15:40:08 dobby ntpd[593]: kernel time sync enabled 6009
Apr 21 23:13:43 dobby ntpd[593]: kernel time sync enabled 2009
Apr 22 00:21:59 dobby ntpd[593]: kernel time sync enabled 6009
Apr 22 09:27:39 dobby ntpd[593]: kernel time sync enabled 2009
Apr 22 13:57:53 dobby ntpd[593]: kernel time sync enabled 6009

If I understand those messages, it looks like ntpd has been restarting  
itself, sometimes 3 times per day.  Is that a correct interpretation?   
Is that normal?  What conditions would lead to such behavior?

Any help welcome.  (Other than "read the ISC documentation" which I  
find impenetrable)

Cheers,

-j


-- 
Jeffrey Goldberg                        http://www.goldmark.org/jeff/




More information about the pool mailing list