[lug] ntp woes again

[Tkil]

>>>>> "MJH" == Michael J Hammel <mjhammel at graphics-muse.org> writes:

MJH> Isn't ntpdate a preferred method for setting my system time?

it's been deprecated.  the new solution is to use the same ntpd binary
as the daemon does, but run it with flags that tell it:

1. exit as soon as the clock has been successfully set

and optionally:

2. set the clock even if it's more than the "sanity limit" away from
   the current time.

the options to do this are explained in the ntpd docs.

doing this with ntpd has the advantage/disadvantage of using the
common configuration and drift files.

for whatever it's worth, i'm not sure i've got my configuration
correctly set up.  i'm still seeing more "time synchronisation lost"
messages in my log than i'm comfortable with.  but it's close enough
for now.

MJH> If this is a firewall issue, shouldn't I be getting messages
MJH> about rejected packets in /var/log/messages?

only if your rejects are logged -- redhat does not do this by
default.  (i've added it to all my reject rules, and mostly it's been
a good thing.)

t.