Steve,
I agree with you about echo requests, to a point. But the host requirements
were written back when a username and password were sufficient for security.
(I remember the good-ol' days of the NSFnet in the early 80's). In those
days, a "firewall" was something made from approved flame-retardant
materials.
Today, IT managers routinely spend more time fighting spam and hackers than
doing anything else. I know the time I've spent on iptables and postfix
rules never fails to remind me of how sad it is that I have to spend any
time at all on them. So in the corporate world, it is common for key
servers to NOT respond to pings.
I did a test with my gateway when I first put it up. I had it running for a
month or so without accepting pings from the Internet. As soon as I allowed
pings, a variety of attacks began, all within a day or so. Luckily, none
were successful.
My solution was to allow pings from within AMPRnet, but not from external
"commercial" Internet machines. That makes it helpful for experimentation
and testing within our ham community, but reduces visibility to the outside.
The problem with that approach is that it requires a bit more care and
configuration knowledge in iptables or whatever firewall solution someone
has in place. For many, that's either something that is too much to deal
with or something they try to do but aren't quite able to get the
configuration right.
Bottom line, I think it would be a mistake to rely on ICMP echo
requests/response for proof that a gateway is there or not. Our local
repeater coordinating body requires coordination holders to update their
record once every two years (or maybe it's three years). It's as simple as
logging in and clicking a button indicating that their information is
current. The database sends automated reminder emails when the update is
due. It has worked well for many years and is not onerous for the
administrative/system staff or the repeater owners. Something similar seems
like the way to go.
Michael
N6MEF
-----Original Message-----
From: 44net-bounces+n6mef=mefox.org(a)hamradio.ucsd.edu
[mailto:44net-bounces+n6mef=mefox.org@hamradio.ucsd.edu] On Behalf Of Steve
Platt
Sent: Wednesday, March 13, 2013 2:25 AM
To: AMPRNet working group
Subject: Re: [44net] Portal registrations
My belief is that systems connected to the Internet should respond to ICMP
Echo Requests, it is/was in the Host Requirements RFC I think. It's also so
useful in an experimental network that I would not want to turn it off.
On the other hand, I strongly believe that allocation of static IP addresses
should not *automatically* depend on whether a system is switched on 24/7,
nor on whether the system responds to a ping. Static addresses need to be
managed manually.
If address space is scarce, allocate it dynamically. That's what DHCP is
for.
Best wishes and thanks to all who keep AMPRnet alive!
G4WSZ
--
Steve Platt