[plug] NTP update problems - dialup

[William Kenworthy]

check out "chrony" - its supposed to be modem friendly. However, I found
a properly setup ntp was better and less trouble than chrony in the long
run.  Chrony had a few problems with the performance patch sets in the
gentoo kernels at one time as well.

Eventually I just set up ntp and let it run - time was always well
synced.  Also, you can kill and restart ntpd by script when the line
comes up (I do this on adsl when it sometimes acts up when I get an ip
change).  Using the correct options you can get ntp to immediately sync
on startup.

My adsl ntp.conf, should work on a modem:
____________________________
tinker panic 0 huffpuff 7200
server 127.127.1.0
fudge 127.127.1.0 stratum 10
server time.uwa.edu.au
server ntp.iinet.net.au iburst burst prefer
server spider.murdoch.edu.au
server lobster.murdoch.edu.au
server oceania.pool.ntp.org
server oceania.pool.ntp.org
server oceania.pool.ntp.org
driftfile /etc/ntp/drift
____________________________

I am on iinet, and their timeservers are quite stable enough for general
use. All the perth servers are far more stable than the pool ones - they
are usually too far away in comparison.  Servers in perth often (but not
always) point to the same master at UWA.  The status is below, you can
see one of the problems with pool servers (dns supplying duplicates).
Its not too bad at the moment, esp as we had a power bump about
~1630/1645, but the pool servers are often bad.  Only keep them in as it
makes the perth ones look good! - should remove them I guess.
Reccomendations I have seen say about 6 servers is good for serious
timekeeping, and 3 for home ones.  My internal machines point to the
main server on the gateway, and an iptables rule snaffles ntp packets
and redirects them to the gateway for guest machines like my laptop.

>ntpq -p

     remote           refid      st t when poll reach   delay   offset
jitter
==============================================================================
 LOCAL(0)        LOCAL(0)        10 l   16   64  377    0.000    0.000
0.002
-elysium.uwa.edu 130.95.156.8     2 u  668 1024  377   23.930    3.900
0.538
*hrsp-core.wa.ii 128.250.36.3     2 u  742 1024  377   25.468    2.163
4.320
+spider.murdoch. 130.95.156.8     2 u  679 1024  377   23.878    2.856
0.193
+lobster.murdoch 130.95.156.8     2 u  671 1024  377   23.877    3.009
0.671
 gen2.ihug.co.nz .INIT.          16 u    - 1024    0    0.000    0.000
4000.00
+gen2.ihug.co.nz 203.167.224.60   2 u  556 1024  377  104.120    2.599
4.163
+gen3.ihug.co.nz 130.217.76.49    2 u  546 1024  377  103.904    2.980
2.186

On Tue, 2005-03-29 at 17:47 +0800, Timothy White wrote:
> Gavin Rogers wrote:
> 
> > At 10:25 AM 29/03/2005, you wrote:
> >
> >> Once again I'm having problems getting synchronisation with a time
> >> server. What I have down is right a small script and put it in