[tor-bugs] #7157 [Tor]: "Low circuit success rate %u/%u for guard %s=%s."

[Tor Bug Tracker & Wiki]

#7157: "Low circuit success rate %u/%u for guard %s=%s."
-----------------------------------------+----------------------------------
 Reporter:  arma                         |          Owner:                    
     Type:  enhancement                  |         Status:  needs_revision    
 Priority:  normal                       |      Milestone:  Tor: 0.2.4.x-final
Component:  Tor                          |        Version:                    
 Keywords:  tor-client, MikePerry201211  |         Parent:                    
   Points:                               |   Actualpoints:  8                 
-----------------------------------------+----------------------------------

Comment(by mikeperry):

 Replying to [comment:12 nickm]:
 > Replying to [comment:11 mikeperry]:
 > > I created #7341 for the timeout issue. Any magical insight on the
 truncation problem? Should we just switch to double-precision floating
 point for the state file counts? Or maybe we should use a real (weighted?)
 moving average and simply store a count and a double rate value, and
 publish some kind of weight in the consensus?
 >
 > Hm.  I think that using doubles would be okay IMO.  It sounds less
 error-prone than most of the other approaches.  Does it sound less error-
 prone to you?

 Less error-prone to get to from what we've got, yes. However, I am still
 tempted to change it to an EWMA of some kind, because that's more typical
 and less crazy than this scaling idea, which I think only makes sense if
 we're staying in the integers. I can update my simulation to try out
 various EWMA and see what I think after that.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/7157#comment:13>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online