[tor-bugs] #1772 [Tor Client]: Counting circuit timeouts when we can't establish any TLS connection

[Tor Bug Tracker & Wiki]

#1772: Counting circuit timeouts when we can't establish any TLS connection
------------------------+---------------------------------------------------
 Reporter:  arma        |       Owner:  mikeperry         
     Type:  defect      |      Status:  assigned          
 Priority:  major       |   Milestone:  Tor: 0.2.2.x-final
Component:  Tor Client  |     Version:                    
 Keywords:              |      Parent:                    
------------------------+---------------------------------------------------

Comment(by arma):

 http://freehaven.net/~arma/tord-bug1772 is the beginning of a Tor debug
 log. When I wake up (and it's been running for hours) I'll post the rest
 of it.

 Interesting points:

 A) Note the frequent timeouts caused by the Pandoras. They're mostly one-
 hop circuits, so in theory they'll go away with the recent patches to
 #1740. I worry though that we'll just mask the problem then.

 B) We should not do the CIRCUIT_PURPOSE_C_MEASURE_TIMEOUT dance if
 cbt->close_ms == cbt->timeout_ms. It just causes more timeouts.

 C) There are a few instances of bug #1980 for you, complete with debug
 logs:
 {{{
 Sep 29 09:51:48.575 [warn] circuit_build_times_network_close(): Bug:
 Circuit somehow completed a hop while the network was not live. Network
 was last live at 2010-09-29 09:50:47, but circuit launched at 2010-09-29
 09:50:47. It's now 2010-09-29 09:51:48.
 }}}

 D) I expect that when it reaches 100 build times, it will produce an
 atrocious cbt. We should think about how to cap the number at something
 realistic. I am guessing that these Pandoras are simply never planning to
 answer me, no matter my timeout.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/1772#comment:9>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online