Roger Oberholtzer wrote:
On Fri, 2009-10-30 at 00:18 +0100, Carlos E. R. wrote:
Exactly what I said would happen. NTP runs "happily" with no network, but it quits, aborts, when it finds the time difference is too big.
It must be the presence of both factors at the same time (no network, difference too big).
Difference too big is the actual reason that ntpd will die. That difference is 1000 secs, btw. AFAIR, SUSE adds local clock to its ntp sources, and thus ntp will have an available time source and will not die if the other time sources are not reachable. (I have my own ntp configs and can not check SUSE's default config.) A big problem at start up of xntpd and ntpd is that the mitigation algorithms are initially crippled in the interest of a fast start, and may not detect a falseticker and will happily sync to it if it is the first server that becomes available. In the SUSE case, the falseticker is the local clock, which may be more than 1000 secs off. Suppose that you have no network at first, and then add one. Then (maybe) four servers come available. After 5 polls, the truechimers will vote the false ticker off the island, and the system will now want to reset to the correct time, but since this is not the first sync, the 1000 second limit will come into play and ntpd will exit. This is an architectural issue; I don't know how to handle that problem for mobile systems. For all kind of configurations, one can find counter scenarios where it doesn't work. The basic premise of ntp is `always online' and doesn't take into account wild straying systems that are not regularly connected (and long enough to sync!) to the network. I happen to leave off local clocks in our ntp client configuration -- not on the ntp servers, but there one has to take care that the local clocks *must* be in a different stratum, otherwise this will create havoc. But that still gives problems in some cases, so YMMV. Configuring ntp properly is a minefield; it looks easy at first and is very difficult in the details. And it doesn't help that the documentation does not care for common cases. HTH nevertheless, Joachim -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Joachim Schrod Email: jschrod@acm.org Roedermark, Germany -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org