Subject: Re: DF bit copying in gif processing
To: None <>
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
List: tech-net
Date: 10/17/2000 10:37:48
In message <>Jason R Thorpe writes
>On Tue, Oct 17, 2000 at 01:10:55PM -0400, Bill Sommerfeld wrote:
> > Except that, in practice, the folks who needs to start doing
> > black-hole discovery are the N zillion web servers sitting behind
> > their own icmp-opaque firewalls....
>So, if an admin has an icmp-opaque firewall in front of his web farm,
>he deserves to lose.  Losing may include turning off PMTU Discovery
>on his web servers.

That's not reality.  The reality is that when some thirdparty does
this, and PMTU is turned on in NetBSD, then NetBSD users lose.  not
the third-party admin.  If those users don't lose with other systems,
then that is legitimately a problem with NetBSD.

Pointing at IETF committee structure and saying "oh, blackhole
discover is a separate job" doesnt' excuse us from doing reasonable
engineeering.  We need to do an adequate job of blackhole discovery.
If we don't, our PMTU is broken, period.