On Tue, Jun 29, 2004 at 03:48:31PM +0200, Corné Beerse wrote: > T. Nifty Hat Mitchell wrote: > >On Thu, Jun 24, 2004 at 08:34:02PM +0200, Eric Tanguy wrote: > > > > > >>In the boot process the ntp synchronisation failed whereas after the > >>time synchronization is all right. Why ? may be because i use > >>firestarter ? > > > > > >If you have friends running Linux/Fedora exchange NTP privileges. > > > >First look at how ntpd is started for you. > > > >In general you will find that the initial connection using ntpdate > >times out for various normal reasons. As a result the return status > >is an error. > > You can chech for this behavoure to do a ping to the ntp-server. If you > ping 3 to 5 times, you will see the first ping takes longer than the > others. Most likely, after this ping (before the ntpd-start) the first ntp > query also succeeds. > > Specially if you call the ntp-server by name, the IP-address is not cashed > jet and hence a dns-query must be made. Since it is at boot time, most > likely all routings and such also need to be setup. > > If you have pinged the ntp-server (even only once!), all routers and > name-caches are filled and the ntp-query can go strait on and will be > substantial faster. > > > > >If you check back +15 min later things will be fine when ntpd has had time > >to make some connections. > > However, if you start ntp only 15 minutes later, it is likely that the > first call still is a little late. > > > > >Note that /etc/init.d/ntpd first starts ntpdate then ntpd. It is not > >uncommon for ntpdate to fail in the early stages of booting a system > >for a list of reasons... > > Not necessarily, newer ntpd-s have ntp-date build in and do that themself. > If it is done separatly, this ntp-date will walk the paths for the next > ntpd-s. > > > > >Do, find a ntp time service local to you. > >Based on the mail headers, perhaps ntp.univ-nantes.fr > > > >or look here... http://www.eecis.udel.edu/~mills/ntp/clock2b.html > > >If you have friends running Linux/Fedora exchange NTP privileges with each > >other! Yes the way that the current scripts start the ntp time daemon might be improved. I expect when ntpdate is eventually removed the change will take place. I spent a little time looking at it a year or so back and the standard pair of redhat time servers seem to be overloaded enough that the initial connections always time out and a RED FAIL was just too common for me. I looked for closer and more reliable time servers. It really does help to exchange keys with other Linux friends and as a group get connected to a pair of near by level one or two hosts depending on how large the group is. If your ISP does not document a NTP time host ask. NTP is such a high quality service that a pair of level 3 servers will provide exceedingly good time of day references. Of interest there is a multicast/broadcast ntp protocol. Opening the firewall to this port might find the service already active. -- T o m M i t c h e l l /dev/null the ultimate in secure storage.
