[tor-bugs] #13239 [Tor]: Maybe we want three preemptive internal circs for hidden services?

Tor Bug Tracker & Wiki blackhole at torproject.org
Fri Jul 10 18:48:05 UTC 2015


#13239: Maybe we want three preemptive internal circs for hidden services?
------------------------+---------------------------------
     Reporter:  arma    |      Owner:
         Type:  defect  |     Status:  new
     Priority:  normal  |  Milestone:  Tor: 0.2.???
    Component:  Tor     |    Version:
   Resolution:          |   Keywords:  SponsorR tor-client
Actual Points:          |  Parent ID:
       Points:          |
------------------------+---------------------------------
Description changed by arma:

Old description:

> {{{
>   /* Fourth, see if we need any more hidden service (client) circuits. */
>   if (rep_hist_get_predicted_internal(now, &hidserv_needs_uptime,
>                                       &hidserv_needs_capacity) &&
>       ((num_uptime_internal<2 && hidserv_needs_uptime) ||
>         num_internal<2)) {
> }}}
>
> Right now we build 2 preemptive internal circuits if we're accessed a
> hidden service lately.
>
> But when we click on a new .onion address, we use three circuits -- one
> to do the hsdesc fetch, one for the rend point, and one for the intro
> point.
>
> You can see it in this walkthrough:
> https://trac.torproject.org/projects/tor/wiki/doc
> /TorControlPortWalkthrough-HS
> where we cannibalize the hsdesc fetch circ, cannibalize the rend point,
> and then have to build the intro point from scratch.
>
> We could raise the above '2' to '3'. I'm not sure if it's worthwhile in
> practice though. Sure would be nice to have some way of answering that
> question.

New description:

 {{{
   /* Fourth, see if we need any more hidden service (client) circuits. */
   if (rep_hist_get_predicted_internal(now, &hidserv_needs_uptime,
                                       &hidserv_needs_capacity) &&
       ((num_uptime_internal<2 && hidserv_needs_uptime) ||
         num_internal<2)) {
 }}}

 Right now we build 2 preemptive internal circuits if we've accessed a
 hidden service lately.

 But when we click on a new .onion address, we use three circuits -- one to
 do the hsdesc fetch, one for the rend point, and one for the intro point.

 You can see it in this walkthrough:
 https://trac.torproject.org/projects/tor/wiki/doc
 /TorControlPortWalkthrough-HS
 where we cannibalize the hsdesc fetch circ, cannibalize the rend point,
 and then have to build the intro point from scratch.

 We could raise the above '2' to '3'. I'm not sure if it's worthwhile in
 practice though. Sure would be nice to have some way of answering that
 question.

--

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/13239#comment:2>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online


More information about the tor-bugs mailing list