On Saturday 10 December 2005 00:33, Tim wrote:
>On Fri, 2005-12-09 at 17:42 -0500, Gene Heskett wrote:
>> One really should be using pool.ntp.org, which is a round robin
>> dynamic assignment, giving ntpd 3 servers to have as a choice of
>> peer.
>
>I've found that using "pool.ntp.org" usually means that I keep on
> using the same set of NTP servers (i.e. no round-robin behaviour).
> I found you needed to be a bit more specific in specifying different
> pool server addresses to get different servers.
>
Its working here, giving ntpd a different set of servers each time its
restarted, and always has.
>e.g. server 0.pool.ntp.org iburst
> server 1.pool.ntp.org iburst
> server 2.pool.ntp.org iburst
>
>If you experiment around, you notice that one of numerical sub-domain
>prefixes is the same as just pool.ntp.org, and they all seem to
> produce the same results for each query (e.g. what's listed as the
> first IP for 0.pool.ntp.org is always listed as the first IP, and
> the NTP client will keep on using the first answer).
>
>And you really need more than three servers, in case you get some
> that don't respond (I frequently find two don't respond) or agree
> with each other (if one is different, how do you know that the other
> two aren't wrong, even if they agree with each other).
>
>As well as the above, more general, pool.ntp.org servers, I added a
>couple of supposedly local ones. Since I'm in Australia, I picked
> the au.pool.ntp.org and nz.pool.ntp.org servers, and my ISP's own
> NTP server, giving me five different servers.
>
>I also wiped out the /etc/ntp/step-tickers and /etc/ntp/ntpservers
> files as they seemed to cause NTP to not work according to how I
> wanted it configured. DHCP assigned NTP server addresses muddy
> things, as well.
As above, that has not been my findings here, although I do not use
dhcp except for my outside address for the dsl connection. The router
serves as a gateway, and it handles the pppoe stuff internally.
Address leases are apparently at least 30 days long with vz.
If you nuked your /etc/ntp/ntpservers file, then all bets are off as
its the ntpd daemon that randomly chooses its 3 peers from that file
at startup. I have added, interspersed with the rest of the named
sites in that file, several more instances of pool.ntp.org so it
stands a pretty good chance of using it for at least 1 of its peers.
But I don't think its using it atm:
[root@coyote netbeans]# ntpq -p
remote refid st t when poll reach delay offset
jitter
==============================================================================
*time.uswo.net 128.10.252.6 2 u 579 1024 377 84.477 0.817
0.138
+rrcs-24-172-8-1 80.64.135.105 3 u 625 1024 377 71.179 6.302
4.245
+ecoca.eed.usv.r 80.96.120.253 2 u 645 1024 377 220.358 0.091
14.791
LOCAL(0) LOCAL(0) 10 l 1 64 377 0.000 0.000
0.001
I don't believe that pool.ntp.org is any of those. But note the offset
of the * entry, its peer chosen to synch to, thats in milliseconds.
Basicly, it will work, if *you* let it. I went thru that phase too.
The only file you should nuke at any time is /var/lib/ntp/drift, do
this when updateing the kernel or the bios and let ntpd regenerate it,
which takes about an hour.
--
Cheers, Gene
People having trouble with vz bouncing email to me should use this
address: <gene.heskett@xxxxxxxxxxxxxxxxx> which bypasses vz's
stupid bounce rules. I do use spamassassin too. :-)
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.
