[tor-bugs] #8235 [Tor]: pathbias_count_build_success()

[Tor Bug Tracker & Wiki]

#8235: pathbias_count_build_success()
-------------------------+--------------------------------------------------
 Reporter:  cypherpunks  |          Owner:  mikeperry          
     Type:  defect       |         Status:  assigned           
 Priority:  major        |      Milestone:  Tor: 0.2.4.x-final 
Component:  Tor          |        Version:  Tor: 0.2.4.10-alpha
 Keywords:  tor-client   |         Parent:                     
   Points:               |   Actualpoints:                     
-------------------------+--------------------------------------------------
Changes (by mikeperry):

  * owner:  => mikeperry
  * status:  new => assigned


Comment:

 FTR, I saw something similar while I was testing the path use bias code in
 0.2.4.9-alpha. There was a source of overcounting present during scaling
 our counts in versions prior to 0.2.4.10-alpha that may have been what
 caused your high counts here. It is possible that the miscounting began
 while you were running those earlier versions.

 If you want to start fresh, stop your tor daemon and run 'grep -v
 "EntryGuardPath" state > state.new ; mv state.new state' in your Tor's
 data directory, then start your tor daemon again. This will clear out the
 old path bias data.

 If it happens again after that, we might need more general information
 about how you're using that Tor daemon (as a hidden service server? with a
 tor controller other than Vidalia? as a relay? with TBB-alpha? with a
 custom client app/script?) so we can try to reproduce here. But man, the
 state machine changes in tor 0.2.4.10 really should make it impossible to
 count the same circuit construction twice, and should have eliminated the
 error source during scaling from 0.2.4.9 and earlier...

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/8235#comment:2>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online