another unusual connection

["john smith" <d1nqf0s4vg9wk5ilq@xxxxxxxxx>]

on Mon, Jan 28, 2008 at 3:33 AM Roger Dingledine wrote:

> Neat. So it was 212.112.242.159 in both cases?

yes

> Please let me know if it happens more (or if other people experience it
> and can provide more details!), and maybe we'll narrow in further.

will do. if there's anything i can do to provide you with more
detailed/specific information then let me know & i will endeavour to
do so.

regards,

john smith

On 1/28/08, Roger Dingledine <arma@xxxxxxx> wrote:
> On Sun, Jan 27, 2008 at 10:42:14PM +0000, john smith wrote:
> > another recurrence of the same type of unusual connection.
> > i include the time the server started in the log below. the connection
> > through 212.112.242.159 persists for a much longer period of time on
> > this occassion (the 'scrubbed' connection did not occur last time).
>
> Neat. So it was 212.112.242.159 in both cases?
>
> New theory: in rare cases, Tor servers (like maximator) lie to directory
> clients about what IP address they appear to have, due to iptables
> confusion or something similar. More specifically, it claims that
> everybody looks like itself. Then Tor servers that don't know their own
> address get suckered into thinking they switched.
>
> If this is actually the bug, I'll have to ponder how to fix it well. We
> could require several places to agree before we think we should switch;
> but that would slow down reaction times considerably. We could only
> believe answers from authorities; but I don't want to preclude better
> load balancing. We could ignore it when we ask a directory mirror at IP
> address X and he says we look like we're coming from IP address X;
> that's probably a good idea, and I should add a check for this. Then we
> can see if that check ever triggers.
>
> Please let me know if it happens more (or if other people experience it
> and can provide more details!), and maybe we'll narrow in further.
>
> Thanks,
> --Roger
>
>