[tor-bugs] #8575 [Stem]: Integ tests segfault python 3

[Tor Bug Tracker & Wiki]

#8575: Integ tests segfault python 3
---------------------+------------------------------------------------------
 Reporter:  atagar   |          Owner:  atagar
     Type:  defect   |         Status:  new   
 Priority:  normal   |      Milestone:        
Component:  Stem     |        Version:        
 Keywords:  testing  |         Parent:        
   Points:           |   Actualpoints:        
---------------------+------------------------------------------------------

Comment(by atagar):

 A potential cause of this is the stem.util.system integ tests which
 include a process renaming integ test. That test gets pretty low level,
 using ctypes to change the present process' name.

 This would be pretty easy to either verify or rule out. Before contacting
 upstream we should do a binary search of our integ tests to narrow them
 down to the subset needed to reproduce this issue. This should quickly
 narrow the haystack quite a bit.

 Another potential issue could be lingering threads. I once spent the
 better part of a month troubleshooting [https://trac.torproject.org/2412
 orphaned threads that caused TorCtl stacktraces while shutting arm]. On
 very rare occasions (0.1%?) I've seen a similar issue with stem's reader
 or event daemons. From my read of the code those threads really *should*
 be getting joined. Unfortunately tracking down such a transient issue has
 been tricky.

 All this said, my money's still on python3 instability. In short we
 should...

 * check that this still occurs with the latest development version of
 python 3.x
 * figure out the set of integration tests we need to run to reproduce this
 * contact upstream and work with them to figure out where the segfault is
 coming from

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/8575#comment:1>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online