[ntp:questions] poll interval - RFC compliance question

Richard B. Gilbert rgilbert88 at comcast.net
Fri Jun 13 19:04:17 UTC 2008


Anton Persson A wrote:
> Hi, 
> 
> we are using the ntp v3 protocol, as defined in RFC1305, however
> I have a question running through my head when I consider the poll
> interval.
> 
> According to RFC 4330: (yes I know, SNTP v4)
> A client SHOULD increase the poll interval using exponential
>        backoff as performance permits and especially if the server does
>        not respond within a reasonable time.
> 
> Is this requirement also valid for ntp v3, I can't quite find it
> but I'm unsure? The reason I'm asking is that we have quite
> a blood thirsty testing squad here and I need to clarify
> if this is a compliance requirement or not.
> 

A little history is in order here!

There have been at least two incidents in which poorly 
designed/implemented NTP or SNTP clients have caused serious problems by 
polling a server at one second intervals.  In one case, many thousands 
of Netgear Routers polled the University of Wisconsin's NTP server at 
one second intervals and brought it to its knees!  See:
http://pages.cs.wisc.edu/~plonka/netgear-sntp/

In another case DLink was the culprit and Poul Henning-Kamp (PHK) was 
the victim.  I believe that there was also a problem with a software 
product named "TARDIS".

I would say that the "requirement" is valid for ANY NTP or SNTP version!
This is especially true if you are going to distribute hardware with 
built in firmware implementing SNTP or NTP.

If you screw this one up, your name will be "mud" forever!  It will be 
many years before the Netgear and Dlink screwups will be forgotten if, 
indeed, they are ever forgotten.




More information about the questions mailing list