[tor-bugs] #8782 [Tor]: Don't give up so easily on your guards if the consensus calls them Running

[Tor Bug Tracker & Wiki]

#8782: Don't give up so easily on your guards if the consensus calls them Running
-------------------------+--------------------------------------------------
 Reporter:  arma         |          Owner:                    
     Type:  enhancement  |         Status:  new               
 Priority:  normal       |      Milestone:  Tor: 0.2.5.x-final
Component:  Tor          |        Version:                    
 Keywords:  tor-client   |         Parent:                    
   Points:               |   Actualpoints:                    
-------------------------+--------------------------------------------------

Comment(by arma):

 In theory this won't induce that many further connections, since it will
 only happen in the case where our guard goes down while we're using it.
 Assuming I guess that guards don't do so that often.

 The trick here will be to check pretty often while avoiding loops where we
 try a new connection, become convinced that he's up, reset our re-check
 timing, realize he's down for the same reason we did last time, repeat.

 We'll want to make a decision about whether we want to go the "when one of
 our guards gets into this state, queue it onto a retry schedule" route, or
 something fancier with the aim of achieving similar goals with fewer
 retries, like "when one of our guards gets into this state, retry all the
 guards currently in this state".

 Speaking of these fancier options, check out the first_contact variable in
 entry_guard_register_connect_status(). Do we already accidentally do
 something like this ticket? Is it enough?

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/8782#comment:2>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online