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

Re: Force exitnode oddness

Those nice people at Privoxy have anticipated the need :)
+filter {hide-tor-exit-notation}
+filter-client-headers
.exit/
It looks like cookies are sent properly even though they are stored under the modified domain name. It also looks like some page requisites (images etc) may be fetched from a different circuit i.e. not respecting the forced exit node. Could be a problem if the page contains absolute URIs. It's an interesting problem. I suppose the only way to be sure is to edit the Tor config for strict-exit-nodes and restart it? 

GD
On 30 Sep 2008, at 03:38, John Brooks wrote:


Tor cares about the hostname it is told to connect to having the
'.exit' suffix - privoxy won't modify that (afaik). What you want to
modify with privoxy is the Host header, so the server you're
connecting to can properly handle virtual hosting. So, you want a
filter rule to modify the Host header and remove '.something.exit'
from the end if present. I'm not a privoxy user myself, so I don't
have any insight on how to actually do that, but I hope that helps.

- John Brooks
On Mon, Sep 29, 2008 at 8:14 PM, Geoff Down <downie@xxxxxxxxxxxxxx> wrote:
I have Privoxy, so that is possible (if anyone has the filter rule already I 
would appreciate it).
But if Privoxy strips the .exit bit, how does Tor then get the information 
about which node to use?
GD
On 30 Sep 2008, at 01:13, coderman wrote:
On Mon, Sep 29, 2008 at 5:07 PM, Geoff Down <downie@xxxxxxxxxxxxxx> wrote: 

...
the hostname is being rejected because it contains the .<node>.exit . Presumably that should have been stripped by my Tor client or the exit 
node.


you actually need a proxy that supports the .exit syntax scrubbing,
like privoxy with appropriate rules, in order to use .exit syntax for
a virtual host domain to work properly.  otherwise, the host header
(and possibly other values, like cookie domain) will be incorrect.

best regards,