[ntp:questions] Does the pool directive automatically imply "preempt"?

E-Mail Sent to this address will be added to the BlackLists Null at BlackList.Anitech-Systems.invalid
Fri Feb 8 23:16:49 UTC 2013


Hal Murray wrote:
>  Terje Mathisen <"terje.mathisen at tmsw.no"> writes:
>> Hal Murray wrote:
>>>> It effectively does imply 'preempt', since a pool of
>>>>  servers (from a pool xyz statement) will be monitored
>>>>  and regularly filtered, replacing the worst servers
>>>>  with the results of a fresh dns lookup.
>>>
>>> Does it replace bad servers or only dead ones?
>>>
>>> I think the latter, but I haven't looked at the code.
>>>
>> If you have sufficient servers (like 10?), I believe it
>>  is supposed to drop the worst two every hour and replace
>>  them by new pool lookups, but like you I haven't actually
>>  read that part of the code.
>
> That's a potentially interesting thing to do, but I don't
>   think it works that way.  I just checked a log file.
>  Nothing added or dropped in the past several days.
>
> Besides, if it did work like that, then we would have to
>  argue about how frequently to drop servers and/or add
>  that parameter to the config file.

It will pull more from the pool, up to 2xMaxClock (20),
 if there are less than MaxClock (10) associations,
 or there are less than MinClock (3) {MinSane?} survivors.

As it goes through the associations,
 it tosses excess over MaxClock (10).


So you can have the case where survivor associations
 drops below MinClock (3),
 and it gets more from the pool, up to 2x MaxClock (20)
 and then it tosses the least desirable till it is
 back down to MaxClock (10), and sometimes in that process
 some that get tossed were pool servers that there before
 it went to get more from the pool, ... and you get pool churn.

-- 
E-Mail Sent to this address <BlackList at Anitech-Systems.com>
  will be added to the BlackLists.



More information about the questions mailing list