[ntp:questions] Pool server gone wild

Rob nomail at example.com
Sat Feb 21 17:53:54 UTC 2015


Roger <invalid at invalid.invalid> wrote:
> On 21 Feb 2015 10:52:40 GMT, Rob <nomail at example.com> wrote:
>
>>Roger <invalid at invalid.invalid> wrote:
>>> On 21 Feb 2015 07:54:50 GMT, Rob <nomail at example.com> wrote:
>>>
>>>>It looks like you have created your own problem.
>>>
>>> What problem are you talking about?
>>
>>Your problem to get enough good servers.
>
> When did I say that?
>
> I observed ntpd continuing to poll a server which was off by
> 100s of milliseconds. Are you saying that ntpd didn't drop the
> server because of not enough good servers? A quick eyeball of
> the relevant peerstats shows that ntpd was using at least six
> good servers plus the one that went wild. Methinks you've come
> to an opinion based on too little information.

You started adding pool commands to add more and more servers,
apparently because you believed there would be not enough to
prune your bad server.

I am amazed that you are using the NTP pool, while at the same
time worrying about a bad server that is off by 100s of milliseconds,
and then still not trust ntpd to reject it.

The NTP pool is for people who want to sync there servers reasonably
well, so the clock widget displays the right time when a human watches it.
When you want guarantees, the NTP pool is not for you.



More information about the questions mailing list