[tor-bugs] #16844 [Tor]: Duplicate directory fetches when on high-latency connection

[Tor Bug Tracker & Wiki]

#16844: Duplicate directory fetches when on high-latency connection
------------------------+------------------------------------
 Reporter:  arma        |          Owner:
     Type:  defect      |         Status:  needs_review
 Priority:  High        |      Milestone:  Tor: 0.2.9.x-final
Component:  Tor         |        Version:
 Severity:  Normal      |     Resolution:
 Keywords:  tor-client  |  Actual Points:
Parent ID:              |         Points:  small/medium
 Reviewer:              |        Sponsor:
------------------------+------------------------------------

Comment (by teor):

 Replying to [comment:15 arma]:
 > Replying to [comment:14 teor]:
 > > Now that we connect to multiple servers simultaneously, we should be
 able to relax the timeout to 30 seconds or something.
 >
 > I think we should not retry begindir requests at all. We now have a
 mechanism for trying a series of directory servers, in a staggered way.
 Let's not layer two confusing mechanisms like that on top of each other.
 The original one that we had is broken. Let's stop using it.

 Oh, I think I have half-killed this timeout during bootstrap already. The
 bootstrap code doesn't retry after the 10 second timeout. It just tries
 another server whenever the schedule comes up. But what I do need to do is
 remove the close-on-timeout behaviour.

 I left the non-bootstrap code using the old mechanism which times out
 after 10 seconds and retries. That's probably ok when a client has a live
 consensus and can afford to try for a few hours to get one. But I'd be
 happy to change that in 0.2.9 as well.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/16844#comment:16>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online