[ntp:questions] Re: ntpdsim problems in ntp-4.2.2.p3
Richard B. Gilbert
rgilbert88 at comcast.net
Thu Sep 7 18:24:01 UTC 2006
Albert J Sawyer wrote:
> Can someone help me get ntpdsim running right? The time will not
> synchronize. At the end of a run with
> ntpdsim -A -C .001 -T 400 -W 1 -c ./ntp.conf
> the last line output is:
> T T+ERR T+ERR+NTP
> 86400.000000 34513621.245209 86400.000000
> By adding -d -d -d to the argument list, and some extra debug output
> messages I found that the clock_select() function always gets 0x400
> failure return from peer_unfit(), which is a Test 11 failure, root
> distance too large.
> A typical line of rawstats has
> 15020 63118.000 18.104.22.168 22.214.171.124 63117.000000000
> 22742575.719312813 22742575.720312811 63117.019083045.
> It seems that the client packet is sent with transmit time from
> ntp_time, but srvr_rply() sets its time stamps from clk_time. Because
> the server is never accepted by clock_select, ntp_time is never adjusted.
> I got this version in downloaded of ntp-4.2.2p3, ran configure.guess to
> create a directory for sparc-sun-solaris2.8, ran configure to create the
> makefile, compiled
> with gnu tools, and created a simple ntp.conf:
> disable kernel
> server time.nist.gov iburst
> logfile /home/ajs/public_html/NTP/ntpdsim/ntplog
> driftfile ./ntp.drift
> statsdir ./ntpstats/
> filegen loopstats file loopstats type day enable
> filegen peerstats file peerstats type day enable
> filegen clockstats file clockstats type day enable
> filegen rawstats file rawstats type day enable
> filegen sysstats file sysstats type day enable
> and ran with:
> ntpdsim $* -A -C .001 -T 400 -W 1 -c ./ntp.conf
> Only peerstats and rawstats are created.
> Al Sawyer
What are you hoping to accomplish with ntpdsim??
Most of us use ntpd and find that it synchronizes our clocks quite well.
I believe ntpdsim is a special purpose tool used by theoreticians and
possibly developers for testing something or other.
Try "ntpd -g".
Unless you are going to be serving time to 100 or more clients, you
should not be using time.nist.gov or any other stratum 1 server. See
the "Rules of Engagement" at
Briefly, most stratum 1 servers have hundreds of thousands if not
millions of clients banging away at them; many with broken
software/firmware that causes them to poll once per second. They tend
to be badly overloaded, to say the least.
More information about the questions