[tor-bugs] #25347 [Core Tor/Tor]: Tor stops building circuits, and doesn't start when it has enough directory information

[Tor Bug Tracker & Wiki]

#25347: Tor stops building circuits, and doesn't start when it has enough directory
information
-------------------------------------------------+-------------------------
 Reporter:  teor                                 |          Owner:  (none)
     Type:  defect                               |         Status:  new
 Priority:  Medium                               |      Milestone:  Tor:
                                                 |  0.3.3.x-final
Component:  Core Tor/Tor                         |        Version:  Tor:
                                                 |  0.3.0.6
 Severity:  Normal                               |     Resolution:
 Keywords:  031-backport, 032-backport,          |  Actual Points:
  033-must, tor-guard, tor-client, tbb-          |
  usability-website, tbb-needs                   |
Parent ID:  #21969                               |         Points:  1
 Reviewer:                                       |        Sponsor:
-------------------------------------------------+-------------------------

Comment (by s7r):

 All 3 guards in the primary set were running Tor versions that do not have
 the DoS patch, so this can be eliminated.

 In regards to the 1/3 primary guard being overworked, I can't prove wrong,
 but:
 - usually this is warned in the client log "your guard X is failing a
 larger amount of circuits than usual. success counts are x/y - this could
 indicate an attack but also mean that the Tor network is overloaded ..." -
 there were no such messages in the client log;

 - why does it not heal by itself even in ~7 hours? Shouldn't it either
 move to another guard or do something, rather than keep running in an
 useless state?

 Some more interesting stuff:
 - checked all 3 guards in the primary set and they were behaving normally,
 running in consensus with good bandwidth values;

 - after starting with a new fresh state file (like starting from scratch),
 Tor selected new primary guards and connected just fine. So I turned it
 off, manually edited the state file and put back at `confirmed_idx=0` the
 same previous guard that was hanging before (with the old state file).
 Started it again, and it was working like nothing happened. Could it be a
 coincidence that for 7 hours that particular guard was overworked and
 failing circuits, and after 5 minutes when I edit the new state file it
 comes back to normal operation? Possible, non-zero chance for this to be
 true, but I wouldn't bet on it.

 teor said on irc that we might be doing something with the state file or
 the mds of the primary guards. I tend to think in the same direction.
 Right now I am trying to reproduce it so I can provide two state files:
 the good working state file and the not working one, with the same primary
 guards maybe they will help us determine what is going in.

 I think this is the same bug reported by cypherpunks on the parent ticket
 and by alec, so it might be the last annoying guard bug.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/25347#comment:6>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online