[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [tor-relays] Thoughts on InspecTor?



On Wed, Feb 01, 2012 at 11:36:57PM -0500, Steve Snyder wrote:
> This application claims to identify bad Tor nodes for the purpose of
> excluding them from use:
> 
> http://xqz3u5drneuzhaeo.onion/users/badtornodes/
> 
> Anyone have any thoughts on this?

In general it is a poor plan to change your routing strategy in a way
that makes you different from most Tor users. If you change it enough,
you end up making a signature for your behavior where an attacker can
say "hey, that's the guy who never uses German relays". The result is
that you harm your anonymity. Nobody has really researched how much harm
comes from how much difference -- and we recommend against doing things
that are poorly understood.

>  The sum of bad-exit-flags (8),

As far as I can tell,
http://xqz3u5drneuzhaeo.onion/users/badtornodes/overview.html#blocking
are all tiny relays so it doesn't much matter whether you badexit them
(you're probably never going to encounter them in practice anyway).

For the question of exit relays in Iran, see
https://trac.torproject.org/projects/tor/ticket/4207
https://trac.torproject.org/projects/tor/ticket/4923

> exit nodes that alter payload (4),

It would be great if people would actually report these to us. I believe
that two of those four are ones we've already noticed and set the badexit
flag on network-wide, and two of those four aren't around anymore.

> and long-term-misconfigured (27)

It would be great if people would report these too, if they are actually
having problems.

In theory, Mike Perry's TorFlow scripts out to be able to automatically
recognize when a relay is buckling under the load and starting to fail
connections. In practice it's not there yet, and he could sure use some
help since we're also asking him to maintain our Firefox fork.

> suggests excluding 39 nodes within the Tor config file.
> 
> Is this reasonable?  Are these exclusions appropriate for relays, or
> for end users, or neither, or both?

I'd go with 'neither'.

This is a community that looked at a relay with the nickname
"NSAFortMeade" and freaked out because it was clearly an NSA-run node. You
can pick your nickname to be whatever you like. If the NSA ran a relay
(which it makes no sense for them to do, since they have already suborned
major telco networks like AT&T so they'd do better to watch *other*
peoples' relays), why the heck would they name it NSAFortMeade? And if
somebody came to you claiming you'd better not use it because omg tin
foil hat, how much should you listen to the rest of their recommendations?

Now, that doesn't mean we don't need help trying to make sure all the
relays are behaving correctly. But a person who sets up a Tor hidden
service, and clearly puts quite a bit of effort into it, but never tries
to contact the Tor developers or the Tor directory authority operators?
Maybe he just needs you to help him communicate with us. :)

The right way to set badexit flags is to do it network-wide so we don't
fragment the set of possible path selection behaviors for clients.

Hope that helps,
--Roger

_______________________________________________
tor-relays mailing list
tor-relays@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays