On Sun, Aug 18, 2013 at 10:44 PM, John Winters <[email protected]> wrote:
> On 18/08/13 17:57, Ajay Garg wrote: > [snip] > >> Just curious though, is it supposed to work this way? I mean, how do the >> >> clients know whether the time they are trying to "fetch" from S1 is >> right or wrong. >> > > If your clients are relying on a single server for the time, then there's > no way they can guard against it maliciously sending the wrong time. OTOH, > if you control the server as well then presumably you aren't going to put a > malicious time server on it. > > Assuming you're running a benign time server process (e.g. ntpd) then it's > up to the server, not the clients, to decide whether the time is right or > wrong. > > I'm not familiar with the nitty-gritty of the protocol, but from > observations when setting up a similar configuration, the server appeared > to exercise self-censorship, and refused to serve time requests until it > had satisfied itself that it had got a good idea of the time from its > upstream peers. > I guess that is in sync with my observations too, wherein the server, by itself, refuses to send the wrong time :D Glad then that I note consistent observations :) > > John > ______________________________**_________________ > pool mailing list > [email protected] > http://lists.ntp.org/listinfo/**pool <http://lists.ntp.org/listinfo/pool> > -- Regards, Ajay
_______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
