On Apr 16, 2007, at 12:04 PM, Jeffrey Goldberg wrote:
>> It's not very helpful to CC: every NTP abuse report to the  
>> timekeepers mailing list, especially for the initial contacts.
>
> OK.  I  didn't know what the convention was.  I only recently  
> started copying to the list after I noticed that someone else did.   
> But that may have been a special case.

Perhaps.  It might be useful to seek more feedback from others on the  
list as to what they'd find useful-- I've already put my two cents  
in, so I won't repeat my position.  :-)

[ ... ]
>> I generally don't worry too much about IPs with less than 1  
>> million queries-- that's about equivalent to downloading a single  
>> big file...
>
> I see my effort as more pedagogical than actually defending my  
> network resources.   So I don't have any particular cut off,  but  
> just go after that top two or three when I go through my logs.

That works for me, too.  You might find it helpful to publish your  
network's policy what acceptable use of your NTP servers actually is  
in terms of a rate on the page you email to people, perhaps along  
with a pointer to your NTP stats page(s).

You might want to put this info at the very top, so that people don't  
have to wander through a description of NTP itself to figure out what  
you consider to be a problem.  Or perhaps link to the NTP.org's  
"rules of engagement" page, or the NTP abuse/vandalism page here:  
http://en.wikipedia.org/wiki/NTP_Vandalism

Anyway, at least at the moment, my servers actually don't have any  
significant abusers, even though I've got about 5000 active clients:

   http://pi.codefab.com/ntpstats/ntp_stats.txt

...when the top consumer is using less than 3% of the aggregate  
bandwidth, all is well.  It's when the top consumer is 20-40% that  
you've got a real abusive client.  Of course, opinions vary-- add  
salt to taste.

-- 
-Chuck

_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to