I used severel IPs (as target, not as source) to telnet the
router,
I didn't pick that up from your first message. How certain are
you that these were all answered by the same single router?
all of them where, _at_the_time_ I used them, in use by a ISP
customer which is located in the same building and visiting my
website. All these IPs I tried are within the ISPs IP pool used
for their customers (most of them ADSL and non in the same
building I think).
I'm finding this a little difficult to picture. If I understand
you correctly, the ISP has a pool of addresses to allocate to
customers (dynamically?). At the time you were testing, a whole
bunch of them had been allocated to a single customer in a single
building. (That right there seems odd, unless they provide that
customer with a dedicated pool -- in which case, those addresses
would be reserved and never assigned to any other customer....)
So, I thought:
- the IP (the router was reachable at) belonged to a PC
somehow _within_ their network
It seems less likely that these addresses are all interfaces to
the same router, than that there is a "port forward" configured which
redirects a telnet connection directed at any of these addresses to
a single place.
- there is no need for the ISP to use their _own_ IPs (as
source IP) when remote administering (but maybe this provides a
basic restriction of source IPs?)
My speculation is that they need to be able to use IPs besides
their own, for whatever reason.
- why must the router be reachable _at_ (not from) some
hundreds IP adresses (of their IP pool)?
If this is due to a port forward, the router may not give them a
choice about that.
Is it common practice to admin a router by telnet from
outside, exposing plain text password? Why not use ssh
somehow? Is there no danger that anybody brute forces
the router password and can disable internet access
for all the ISPs customers?
Certainly SSH would be a better choice, IF the router supports it.
While it's true that the Telnet password can be sniffed, they may
argue that the people who have access to sniff it with any real hope
of success are their own staff, many of whom probably already know it
in order to do their jobs.
The disconnect after three wrong tries makes it somwhat challenging
to brute-force this password. The prompts look rather like those
issued by many Cisco routers, on which that first password grants
only limited read-only access and a second password may need to be
broken to get more complete access.
Can this router be locked down more securely? If not the current
model, there are routers on the market which certainly could. Is
the current configuration so easily exploitable that it constitutes
an urgent threat to the ISP and its customers? Possibly, but the
ISP doesn't seem to think so, and they *could* be right. I don't
think you've quite proven them wrong yet, although you've made a
pretty good start.
David Gillett
