[tor-bugs] #6341 [Tor Relay]: connection_or_flush_from_first_active_circuit() does wrong thing when ewma_enabled

[Tor Bug Tracker & Wiki]

#6341: connection_or_flush_from_first_active_circuit() does wrong thing when
ewma_enabled
-----------------------+----------------------------------------------------
 Reporter:  arma       |          Owner:                    
     Type:  defect     |         Status:  needs_review      
 Priority:  normal     |      Milestone:  Tor: 0.2.3.x-final
Component:  Tor Relay  |        Version:                    
 Keywords:             |         Parent:                    
   Points:             |   Actualpoints:                    
-----------------------+----------------------------------------------------

Comment(by arma):

 Replying to [comment:21 robgjansen]:
 > it looks like some crazy shit happened in the middle graph, when the
 patch was not applied. Was this somewhat expected?

 The patch was thought to fix some behavior. We weren't sure if fixing the
 behavior would make a noticeable change. It looks like it is a significant
 change.

 Let me see if I'm interpreting the results correctly. The graphs show that
 ewma-unpatched outperforms the other two -- their requests get to the
 first byte quicker and also get to the last byte quicker. But the data in
 your comment show that far fewer requests are made. Is it fair to say that
 the clients in the ewma-unpatched case put less load on the network
 because they're not making as many requests total? Is this 'not making as
 many requests' correlated to 'not finishing some of their requests in a
 timely fashion'?

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/6341#comment:22>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online