[ntp:questions] Re: NTP sync problems
Robert.Rati at motorola.com
Tue Dec 14 21:02:27 UTC 2004
Richard B. Gilbert wrote:
> Robert Rati wrote:
>> I'm running NTP 4.2 on a Linux system which syncs off of 2 other Linux
>> boxes (One Debian, other Redhat) running ntp (one is ntp 4.1, other is
>> ntp 4.1.2) and I'm having no problems syncing.
>> However, when I use the same client in a different environment which
>> syncs off of a Sun box running NTP 4.2 or a SUSE Linux box running NTP
>> 4.1, I can't sync. When I run ntpq, I see this:
>> [root@ root]# ntpq -pn
>> remote refid st t when poll reach delay offset
>> *127.127.1.0 LOCAL(0) 10 l 31 64 377 0.000 0.000
>> x<IP> .LCL. 1 u 35 64 377 0.272 -33.321
>> x<IP> LOCAL(0) 11 u 44 64 377 0.416 735934.
>> The "x" status means that it is a "designated falseticker by the
>> intersection algorithm". What does this mean? Is this a client or
>> server configuration issue? Can anyone give me some tips on how to
>> debug this situation?
> You have configured two servers which have wildly different notions of
> what time it is. Both servers are serving their unsynchronized local
> Two servers is the worst possible configuration! Ntpd knows for certain
> that at least one of these two is "insane" and has every reason to
> suspect that both are.
Unfortunately, the network topology I must reside within requires two
NTP servers (and only two). In the data I sent, while both servers have
wildly different ideas of time, shouldn't NTP sync against the server
with the lower stratum? If NTP is syncing against two servers with the
same stratum and different times, I can see that NTP wouldn't know what
to do. However, wouldn't NTP sync against the lower stratum server
since there is only one server at that stratum? Or does the time server
sanity checking that NTP uses ignore stratum?
More information about the questions