[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:  new               
 Priority:  normal     |      Milestone:  Tor: 0.2.3.x-final
Component:  Tor Relay  |        Version:                    
 Keywords:             |         Parent:                    
   Points:             |   Actualpoints:                    
-----------------------+----------------------------------------------------

Comment(by robgjansen):

 Replying to [comment:58 arma]:

 > The legend in 'network data read/written over time' is particularly
 poorly placed: what happens to the blue line underneath it? :)

 There's not particularly much room for the legend in the data-over-time
 plots, and its not worth the effort to adjust the legend for every
 simulation run I do. I guess I could place the legend in the upper left
 for the "Read" graph and in the lower right for the "Write" graph...

 >
 > I guess my main question from considering these graphs is: what is the
 variance on these results? If you run with the same network but a
 different random seed, do you get the same general results or are they
 different by a bit? It looks like for the topic at hand, 'different by a
 bit' could easily swing things from 'helps to hurts' or vice versa.

 Sounds like a good discussion for #5398.

 > Does that mean we should close #6341 (now that we fixed the major Shadow
 bug) and move on to simulating something that we think ought to make a
 huge difference?

 Sounds like a good plan.

 Regarding bootstrapping:
 [https://github.com/shadow/shadow/commit/3652ead7192b1764dac1448d83947a787b203db0
 this Shadow patch] has improved things further. Turns out our attempt to
 auto-tune TCP was creating tiny [virtual-]kernel buffers when the end-to-
 end bandwidth-delay product was tiny. After a quick test, the remaining
 small percentage of relays now finish bootstrapping at the same time as
 the others.

 I thought and still wonder if our original auto-tuning was correct. I'm
 afraid its not clear to me what the 'correct' auto-tuning functionality
 is, and I'd rather not browse the kernel source to find out...

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/6341#comment:60>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online