In article <Ej7Wc.38782$Fg5.11673@attbi_s53>,
William Warren <(E-Mail Removed)> wrote:
:> I further seem to recall that Comcast
:> is one of the cable companies that has a "no servers" policy and
:> that they attempt to enforce that policy by having their equipment
:> "scan" (attempt to open) some of the common server ports on all the
:> user machines. If my recollections are correct, then the packets could
:> represent Comcast itself checking to see whether you are running servers.
:That's very unlikely: no "server" is going to respond to a "10" IP address,
:since each customer gets a routable IP address and would only answer probes
:sent to the routable address, not the "10" address assigned to the cable
:modem.
As I recall, at least one of the large cable companies in the USA
assigns 10.* addresses as the customer IP addresses, doing NAT translation
at the network edge. So 10.116.73.53 (for example) could -be- the
outside IP that was assigned through DHCP. If you aren't running a server
[if you aren't allowed to run a server by your ISP contract either],
then you are only making outward connections [with the exception
of a few protocols such as ftp and SIP, which there are "fixup"s for]
and it doesn't matter to you, the end user, that you are using a 10.*
IP that is being translated at the edge of the network.
Besides, if you check the original posting, you will find that the
logs shown by the OP only show one IP address, and that IP address is
obviously the source IP of the packet that was blocked (and no
ports were shown). If the OP was in fact running a server, then the
server software -would- reply to the 10.* packet unless the OP had
taken the time to specifically block those packets.
There is absolutely NOTHING magic at the operating system level
of any OS I am familiar with, that requires extra work or
extra privileges or magic attributes in order to receive packets
sourced from one of the RFC1918 or ARAN reserved IP ranges. If the
packet makes it as far as your equipment, your equipment is going
to receive the packet unless you have those packets specifically
blocked, and it is going to respond if appropriate. If your ISP is
doing it's job, then that response packet is not going to leave the
ISP's network, and so isn't going to make it "back" to wherever the
originator is, but that's a different matter.
:> Thus, a 10.* packet could have originated anywhere in the world that
:> chooses to disregard the "do not let them leak out" clause, and make it
:> over to you through a chain of providers who do not impliment the
:> "please do not let them leak in either" clause.
:I don't think that's a realistic scenario. The "chain of providers" would
:have to have agreements, and router tables, in place to make it possible,
:and I'd be very surprised if any ISP would expose their network to that kind

f unaccountable traffic and the risk it represents.
It *is* a realistic scenario. Follow some of the news.*abuse
newsgroups sometimes. I have seen a number of complaints from people
saying that their providers are letting in 10.* packets, and have taken
it up the chain with their provider only to be told that the provider
has NO intention of changing the situation.
Remember, 3/8 thru 15/8 except 10/8 are all assigned and routable
(1/8 and 2/8 are reserved), so it would be a temptation to peer
using a single /4 CIDR instead of a dozen /8's.
Some of the companies have given the excuse that their peering
agreements don't -allow- them to block any incoming packets.
[Such a situation could have evolved through failover arrangements
in which one ISP agree's to take on all of another ISP's traffic
in case of emergency -- and if the second ISP allowed 10.* within
it's network boundaries, then the 10.* then effectively leaks
over to the other ISP...]
news.*abuse has had a number of people crusading to get ISP's to
block incoming 10.* packets at their network edge, but the problem
persists. Some of the ISPs are actively resisting (follow the money!).
You'd think they'd be egar to block 10.* to cut down on the number
of anonymous virus triggers and such, but many of them are NOT co-operating.
A 10.* source'd packet you receive really might have come from nearly
anywhere, not just within your own ISP's network. It's a serious
problem that some people have spent serious amounts of time to try
to get the ISPs to fix, and the battle is definitely not won yet.
--
Studies show that the average reader ignores 106% of all statistics
they see in .signatures.