Setting up NTP for multicast client service

[james edwards]

On 8/12/06, Philip Prindeville <philipp_subx at redfish-solutions.com> wrote:
>
>
> Ok, well, so I can be sure that it's the broken build and not my config,
> what
> should the config look like ideally?
>
> -Philip



If the router that is offering multicast NTP time service has an IP of
10.1.2.3
this would be the minimum config for a client. You may have to define the
source
interface for NTP on a Cisco router:

multicastclient
restrict default ignore
restrict 224.0.1.1
restrict 10.1.2.3
server 127.127.1.0              # local clock (LCL)
fudge  127.127.1.0 stratum 10   # LCL is unsynchdronized
driftfile /etc/ntp/drift


I think it is easier to debug this by commenting out all "restrict" lines
and seeing if it works.
If it does, your problems are with the ACL's. If it does not work the
problem is elsewhere.

The lines for "server 127.127.1.0" and "fudge  127.127.1.0 stratum 10" allow
NTP to take its time from localhost, if outside sources cannot be queried.
/etc/ntp/ needs to be writable by user ntp. In RH's ntpd, ntpd drops to user
ntp after it starts. I find this file is often owned by root in RH, so you
will never get a drift file written and long term time keeping will not be
that good.

NTPd requites 4 separate sources for time so its "false ticker" algorithms
will operate, allowing one source to go bad. 3 sources are required to offer
long term accurate time.




james
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/fedora-list/attachments/20060812/ff0ef263/attachment-0001.htm>