Re: Single nodes (was same first hops)

2008-10-10 Thread Roger Dingledine
On Fri, Oct 10, 2008 at 09:46:37PM +0100, Geoff Down wrote:
> I don't think I've ever seen the roadmap.

You can find two of them in
https://svn.torproject.org/svn/tor/trunk/doc/design-paper/

Alas, they're not exactly well fleshed out. I've been working on a newer
version that captures more of what we've been up to lately and what we
hope to be up to in the next several years. Stay tuned.

As for the TODO list, take a look at
https://svn.torproject.org/svn/tor/trunk/doc/TODO.021
and
https://svn.torproject.org/svn/tor/trunk/doc/TODO.future

--Roger



Re: Single nodes (was same first hops)

2008-10-10 Thread Geoff Down


On 10 Oct 2008, at 20:36, Roger Dingledine wrote:


On Fri, Oct 10, 2008 at 08:21:54PM +0100, Geoff Down wrote:

They weren't my usual entry nodes, no. It's a mystery.

It's just happened again - 12 single nodes as well as my 3 normal 
entry

nodes and their circuits.
Is there some level of logging I should have on to capture what's 
going

on?


Tor makes its directory fetches using one-hop circuits. It started 
doing

this in 0.2.0.22-rc:
  - Enable encrypted directory connections by default for non-relays,
so censor tools that block Tor directory connections based on their
plaintext patterns will no longer work. This means Tor works in
certain censored countries by default again.

Vidalia shows all the circuits, because it can't really distinguish 
what

you (or your Tor) are planning to use the circuit for.

Down the road, we may switch it so it makes these one-hop circuits to 
your

entry guards. No point revealing your existence to any more relays than
you have to. The phrase for this new design would be "directory 
guards";

you can see it scattered about the TODO file and roadmaps.

--Roger



Aha! Thanks Roger. I went straight from 0.2.0.19 to 0.2.0.31 when I 
downloaded the latest Tor-Privoxy-Vidalia bundle, so if there was 
anything in the 0.2.0.22 release notes I missed it.

I don't think I've ever seen the roadmap.
GD



Re: Single nodes (was same first hops)

2008-10-10 Thread Roger Dingledine
On Fri, Oct 10, 2008 at 08:21:54PM +0100, Geoff Down wrote:
> >They weren't my usual entry nodes, no. It's a mystery.
> >
> It's just happened again - 12 single nodes as well as my 3 normal entry 
> nodes and their circuits.
> Is there some level of logging I should have on to capture what's going 
> on?

Tor makes its directory fetches using one-hop circuits. It started doing
this in 0.2.0.22-rc:
  - Enable encrypted directory connections by default for non-relays,
so censor tools that block Tor directory connections based on their
plaintext patterns will no longer work. This means Tor works in
certain censored countries by default again.

Vidalia shows all the circuits, because it can't really distinguish what
you (or your Tor) are planning to use the circuit for.

Down the road, we may switch it so it makes these one-hop circuits to your
entry guards. No point revealing your existence to any more relays than
you have to. The phrase for this new design would be "directory guards";
you can see it scattered about the TODO file and roadmaps.

--Roger



Single nodes (was same first hops)

2008-10-10 Thread Geoff Down


On 10 Oct 2008, at 07:25, Geoff Down wrote:



On 10 Oct 2008, at 03:40, Scott Bennett wrote:

 On Thu, 9 Oct 2008 19:23:48 +0100 Geoff Down 
<[EMAIL PROTECTED]>

wrote:


Interestingly, I had about 6 single nodes showing on the Vidalia
network map yesterday, whilst my traffic was going via a normal 
3-node

circuit and another 3-node circuit was in preparation.
The single nodes disappeared after 20 minutes or so.

 Were those nodes your entry guards by any chance?  Although tor 
initially
tries to build a few (3?) circuits, once they have expired and no 
longer have
any active streams in them, they get torn down *except* for the links 
between
your client and the entry guard nodes.  This not only improves 
security, but
also means that a new circuit already has the first hop connected 
when tor

goes to build that new circuit.
 Of course, that doesn't explain why those links disappeared 
after about

20 minutes, and right offhand, no other explanation comes to mind.



They weren't my usual entry nodes, no. It's a mystery.


It's just happened again - 12 single nodes as well as my 3 normal entry 
nodes and their circuits.
Is there some level of logging I should have on to capture what's going 
on?

GD