[tor-bugs] #21969 [Core Tor/Tor]: We're missing descriptors for some of our primary entry guards

[Tor Bug Tracker & Wiki]

#21969: We're missing descriptors for some of our primary entry guards
-----------------------------------+------------------------------------
 Reporter:  asn                    |          Owner:
     Type:  defect                 |         Status:  new
 Priority:  High                   |      Milestone:  Tor: 0.3.1.x-final
Component:  Core Tor/Tor           |        Version:
 Severity:  Normal                 |     Resolution:
 Keywords:  tor-guard, tor-bridge  |  Actual Points:
Parent ID:                         |         Points:  1.5
 Reviewer:                         |        Sponsor:  SponsorU
-----------------------------------+------------------------------------

Comment (by s7r):

 asn: you could be right with the second theory. Maybe the descriptors that
 need to be fetched are not even available, because the guard is not
 running in current consensus. This is based on a log analysis I just did
 now:

 On a period of ~1 hour and 15 minutes (20:40 - 21:55), it starts like this
 (this instance is using default Guard context, no bridges):

 {{{
 May 11 20:40:54 electrum Tor[7669]: Our directory information is no longer
 up-to-date enough to build circuits: We're missing descriptors for some of
 our primary entry guards
 May 11 20:40:54 electrum Tor[7669]: I learned some more directory
 information, but not enough to build a circuit: We're missing descriptors
 for some of our primary entry guards
 }}}

 Immediately after, over 500 messages like this one (with 1-3 seconds time
 between them):
 {{{
 May 11 20:41:01 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 20:41:04 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 20:41:07 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 20:41:10 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 20:41:44 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 20:41:47 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 20:41:49 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 20:41:51 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 }}}
 [...] so on so on until:
 {{{
 May 11 21:54:55 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 21:54:57 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 21:54:59 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 21:55:04 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 21:55:05 electrum Tor[7669]: Giving up launching first hop of
 circuit to rendezvous point [scrubbed] for service yzgq5m227nhsi33j.
 May 11 21:55:53 electrum Tor[7669]: We now have enough directory
 information to build circuits.
 May 11 21:55:54 electrum Tor[7669]: Tor has successfully opened a circuit.
 Looks like client functionality is working.
 May 11 21:55:54 electrum Tor[7669]: Tor has successfully opened a circuit.
 Looks like client functionality is working.
 }}}

 And it stopped complaining about failing the first hop of a rendezvous
 circuit. First hop is obviously the guard.

 There was a clear cutoff of over 1 hour in this case, just because one of
 the descriptors was missing. I don't know if the first primary guard was
 in that consensus as running or not.

 To take an informed decision, let's log some more information in a custom
 branch:
 - all primary guards in their order (first, second, etc)
 - the primary guard(s) for which we are missing descriptors only
 - the status of all primary guards in the current consensus (running / not
 running)
 - attempts to fetch the missing descriptors and status of this operation
 (success / failure).

 Maybe something else?

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/21969#comment:13>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online