[tor-bugs] #1952 [Tor Relay]: Bw Weights error 1 for case Case 3b

Tor Bug Tracker & Wiki torproject-admin at torproject.org
Sat Sep 25 15:22:30 UTC 2010


#1952: Bw Weights error 1 for case Case 3b
-----------------------+----------------------------------------------------
 Reporter:  arma       |       Owner:  mikeperry         
     Type:  defect     |      Status:  new               
 Priority:  normal     |   Milestone:  Tor: 0.2.2.x-final
Component:  Tor Relay  |     Version:                    
 Keywords:             |      Parent:                    
-----------------------+----------------------------------------------------

Comment(by mikeperry):

 Ok. I think I've got this bug solved. However, I have two equally valid
 solutions for case 3b, when Exits are scarce but Guards are plentiful.

 One solution causes nodes flagged with Exit+Guard to never be used as
 guard nodes, but still allows both Exit and Exit+Guard nodes to be chosen
 as middle nodes.

 The other solution causes Exit flagged nodes to never be chosen as middle
 nodes.

 The first solution has the advantage that Exit+Guard flagged nodes will
 experience less socket pressure, since they won't have to deal with both
 incoming and outgoing connections. However, this isn't really true,
 because sometimes the network moves to case 2B, which would allow
 Exit+Guard to be guards... The number of clients should still be lower,
 though.

 This solution also allows us to change bridge users to use Guards and
 Exits+Guards as their middle hop, without that hop instantly realizing
 that it is being connected to by a bridge. However, the second solution
 would still allow this, but we would have to code the bridge clients to
 only use Exit+Guard flagged nodes as their exits while this case 3b was in
 effect.

 So which one do we pick, first or second?

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/1952#comment:3>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online


More information about the tor-bugs mailing list