[tor-bugs] #872 [Tor Client]: Endless loop of requests if broken resolves.

Tor Bug Tracker & Wiki torproject-admin at torproject.org
Thu Nov 11 16:56:45 UTC 2010


#872: Endless loop of requests if broken resolves.
------------------------------+---------------------------------------------
 Reporter:  rovv              |         Type:  defect    
   Status:  new               |     Priority:  minor     
Milestone:  Tor: unspecified  |    Component:  Tor Client
  Version:  0.2.0.31          |   Resolution:  None      
 Keywords:                    |       Parent:            
------------------------------+---------------------------------------------
Changes (by nickm):

  * milestone:  post 0.2.1.x => Tor: unspecified


Old description:

> Client to ignore all DNS answers with local addresses by default (v
> 0.2.0.32).
> However, it lead connection_edge_process_end_not_open() to
> endless loop of connections attempt (connection_ap_detach_retriable())
> via the same exit, if relay cell (END_STREAM_REASON_EXITPOLICY)
> contains local addresses.
> Attached some ideas of fix.
>

> [Automatically added by flyspray2trac: Operating System: All]

New description:

 Client to ignore all DNS answers with local addresses by default (v
 0.2.0.32).
 However, it lead connection_edge_process_end_not_open() to
 endless loop of connections attempt (connection_ap_detach_retriable())
 via the same exit, if relay cell (END_STREAM_REASON_EXITPOLICY)
 contains local addresses.
 Attached some ideas of fix.


 [Automatically added by flyspray2trac: Operating System: All]

--

Comment:

 So I no longer remembered  what exactly the "right thing" is supposed to
 be.  Let me try to summarize:

    * When a stream fails, or a resolve fails or gives us a weird answer,
 we should maybe retry on a different exit.  Right now we just close the
 stream.
    * We do not want to mark the whole circuit as invalid, or else any user
 who types "localhost" will cycle through circuits till they run out of
 retries.  We don't want to mark the exit as invalid, or any user who types
 "localhost" will soon run out of exits.

 The current close-the-stream behavior does not seem to be causing problems
 in practice AFAICT.  So I'm going to move this to the "unspecified"
 milestone.  We should revisit that decision if this someday turns out to
 cause problems.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/872#comment:13>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online


More information about the tor-bugs mailing list