Slashdot Mirror
RoadRunner Intercepting Domain Typos
shaunco writes "Sometime around midnight on February 26th (at least for the SoCal users), TimeWarner's RoadRunner service started intercepting failed DNS requests, redirecting them to RoadRunner's own search and advertising platform. To see if this has been enabled in your area, try visiting {some random string}.com in your Web browser. This feature subverts user preferences set within browsers, which allow the user to select which search engine receives their typos and invalid domains. RoadRunner users can disable this function — or they can just use OpenDNS. Here is an example RoadRunner results page.
This has actually been going on for a few weeks now for New York area customers. However, there is an opt-out option that comes up on the page that comes up. I'm not quite sure how it tracks those opt-outs (by ip address perhaps?), as I didn't delve into it too deeply.
How ironic... someone registered www.jkshdfkljh23sadf.com as a parked domain. Wow these ppl need help.
The problem here is that what TW is doing breaks DNS. By the RFCs, when I try to resolve a name that doesn't exist, I'm supposed to get an NX "record does not exist" result. What I get instead is an affirmative A record "name exists at this address" response. What happens at the browser level is irrelevant, TW's DNS system has already lied about the state of the DNS records associated with a given domain. This badly breaks a lot of things that aren't browsers that use HTTP and depend on correct NX responses to tell them when the server they're trying to talk to doesn't exist.
As long as TW doesn't block direct use of non-TW DNS servers this can be worked around. If they start blocking that access, or redirecting all DNS traffic to their servers, then we've got a major problem on our hands.
FAIL for failing to understand how DNS works... Your statement is only true if you are running a caching server. No reason why bind can't do its own lookup. You lose out on the cache benefits of a larger DNS server, but don't have to rely on anything other than the roots.
There was. What TW's doing is more pernicious, though. When NetSol was doing it, they were returning the A records directly from their first-level nameservers. BIND's no-delegation option can deal with that, because those first-level nameservers aren't supposed to be returning A records and BIND can translate those response into proper NX responses. With TW, since their DNS servers are supposed to be returning A records, there's no way to tell whether a particular affirmative response is valid or invalid. The only way to fix the problem is to cut TW's servers out of the loop entirely. All well and good, until of course TW either starts blocking all traffic to port 53 that's not to their DNS servers (like they do with outbound to port 25 now) or silently redirecting all DNS queries to their servers. Note that both of these are trivial, my own firewall has (commented-out) rules for both and neither takes more than about 3 lines.
They're tracking by the cable modem's MAC address. There's a page explaining this (and how it's insecure) here:
http://rgov.org/road-runners-dns-wildcard
OpenDNS has a blog post explaining why they're doing that: http://blog.opendns.com/2007/05/22/google-turns-the-page
The plot thickens. Have a look at this OpenDNS blog entry which explains the rationale for the Google interception. At least it's a plausible justification, though I don't have a Dell and I'd prefer my Googling to go straight to the source without intermediaries, so I'm keeping OpenDNS off.
"Patriotism is your conviction that this country is superior to all other countries because you were born in it." -- GBS