[tor-bugs] #21394 [Core Tor/Tor]: connection timeouts are affecting Tor Browser usability

[Tor Bug Tracker & Wiki]

#21394: connection timeouts are affecting Tor Browser usability
-------------------------------------------------+-------------------------
 Reporter:  arthuredelstein                      |          Owner:  (none)
     Type:  defect                               |         Status:  new
 Priority:  Very High                            |      Milestone:  Tor:
                                                 |  0.3.3.x-final
Component:  Core Tor/Tor                         |        Version:
 Severity:  Normal                               |     Resolution:
 Keywords:  tbb-performance, tbb-usability,      |  Actual Points:
  performance, tbb-needs                         |
Parent ID:                                       |         Points:
 Reviewer:                                       |        Sponsor:
-------------------------------------------------+-------------------------

Comment (by teor):

 Replying to [comment:28 arthuredelstein]:
 > ...
 > One thing I noticed is that bad relays have higher than average
 consensus weights. Here are the average consensus weights for relays with
 different DNS behaviors:
 > * Bad (always timed out): 38612
 > * Flaky (sometimes timed out): 10440
 > * Good (never timed out): 7042
 >
 > That trend explains why I see higher timeout frequency with Tor Browser
 than with my DNS timeout survey script.

 This isn't surprising to me, but it's good to have some data.

 I've noticed that guard relays I run in particular locations (east coast
 US, west coast EU) are overloaded, and others have told me that their
 exits in similar locations are also overloaded.

 If you want to address this issue, we have to either:
 * penalise timeouts more on bandwidth authorities (we need a test
 environment to make changes like this), or
 * distribute bandwidth authority clients and servers outside EU/US (we're
 working on this)

 We could also find and map overloaded relays, or relays with higher than
 average consensus weight to bandwidth ratios. That would help us find out
 if our changes are making a difference. I opened #24045 for this.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/21394#comment:29>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online