As I explained in my first reply, I was led to believe that it is wrong to have the local clock in the configuration. Since then I've been running ntp without the local clock line, and it's been fine since.
I'm not saying that this is your problem, but I suspect that this may not be helping the situation - especially since it appears that ntpd has ruled out the other peers as possible synchronisation sources.
it out now, and ntpd restarted. We'll see if it (ntpq -p) even bothers to report LOCAL now. Nope.
It won't do. ntpq -p reports the _peers_ known to the server. Obviously, if you remove the local peer, it won't be shown in that output anymore. Moreover, think whether it is correct to setup a peering between your local clock and your local clock.
21 Dec 15:22:47 ntpd[9198]: logging to file /var/log/ntp.log
21 Dec 15:22:47 ntpd[9198]: ntpd 4.2.0a@xxxxxxxx Fri Aug 26 04:27:20 EDT 2005 (1)
21 Dec 15:22:47 ntpd[9198]: precision = 1.000 usec
21 Dec 15:22:47 ntpd[9198]: Listening on interface wildcard, 0.0.0.0#123
21 Dec 15:22:47 ntpd[9198]: Listening on interface lo, 127.0.0.1#123
21 Dec 15:22:47 ntpd[9198]: Listening on interface eth0, 192.168.71.3#123
21 Dec 15:22:47 ntpd[9198]: kernel time sync status 0040
21 Dec 15:22:47 ntpd[9198]: frequency initialized 3.712 PPM from / var/lib/ntp/drift
21 Dec 15:27:06 ntpd[9198]: synchronized to 128.6.213.15, stratum 3
21 Dec 15:27:06 ntpd[9198]: kernel time sync disabled 0041
21 Dec 15:31:21 ntpd[9198]: synchronized to 192.36.143.151, stratum 1
21 Dec 15:32:29 ntpd[9198]: kernel time sync enabled 0001
So the kernel's timekeeping transitioned from unsynchronised to PLL mode, to PLL + synchronised. Great, ntpd has adjusted the kernels timekeeping in an attempt to keep it synchronised.