[ntp:questions] Re: ntpd looses sync, I must always restart: please help

Maarten Wiltink maarten at kittensandcats.net
Thu Oct 30 23:43:22 UTC 2003


"Luca" <luca at pacor.de> wrote in message
news:85d8a793.0310301100.ce76e6c at posting.google.com...

> PIII/500, 128MB RAM, SuSE Linux 8.2 (Kernel 2.4.20), DSL Connection,
> Provider=T-Online (automatic line-cut after maximum 24 hours), dial on
> demand activated.
>
> Now, my problem is that I must restart the daemon at least once a day
> because I cannot see any server anymore.

Renewing the associations would suffice. Jan Ceuleers posted a script
for doing this last September 11th. (I have the script saved. But not
the message-id. Google knows.)


>                                          I also tried to set up many
> servers and peers

Indeed. Quite ridiculously many, if I may say so. Conspicuously absent
from your endless list is ntp1.t-online.de.

Also, you're trying to _peer_ with half of them, rather than having them
serve you. Take my word for it, they won't take your word for what time
it is. The purpose of peering is to have a two-way connection in case
one of the servers loses its reference clock.

Note the "reference clock". You don't have one. They don't _need_ your
time; it's not going to be better than anything they have.


>                   (I decided to be stratus 2, is that wrong?)

Yes, that's wrong, too. Uplifting, all this, isn't it? First of all, the
word is "stratum". Stratus is a cloud formation; liguistically, layered
clouds, practically, it's the default name for all clouds that aren't
particularly hazy (cirrus) or towering (cumulus).

But the real error is making your unsynchronised local clock stratum 2,
or even 5. Don't fool yourself that you _are_ synchronising it; this
clock only comes into effect when contact with all real time sources has
been lost and it really is unsynchronised.

Stratum 10 would be appropriate. It is still useable, but unlikely to be
used except when really nothing else is available. Which is precisely true,
respectively precisely what you want to be true, of a free-running crystal
oscillator.

Groetjes,
Maarten Wiltink





More information about the questions mailing list