[tor-project] Fallback Directory Monitoring

[Damian Johnson]

Hi Tim, hi Nick. As requested on ticket #18177 [1] we now have a new
DocTor monitor [2] that checks the reachability of our fallback
directories. A notification that looks like the following is sent to
the three of us and #tor-bots if over 25% of the fallbacks are
unreachable...

------------------------------------------------------------

29/100 (29%) fallback directories have become slow or unresponsive...

  * 9504CB22EEB25D344DE63CB7A6F2C46F895C3686 => IPv6 ORPort is
unreachable (2a03:b0c0:3:d0::2ed:7001:9050)
  * AEA43CB1E47BE5F8051711B2BF01683DB1568E05 => IPv6 ORPort is
unreachable (2001:41d0:a:74a::1:443)
  * 8B7F47AE1A5D954A3E58ACDE0865D09DBA5B738D => DirPort is unreachable
(178.217.184.32:9030)
  ... etc...

------------------------------------------------------------

A fallback is defined as unavailable if...

* We can't connect to its ORPort.
* We can't connect to its DirPort.
* We can't connect to its IPv6 ORPort (if it has one).
* It takes longer than 15 seconds to provide the consensus (download
times are usually ~3s).

The cron runs once per day. Happy to also send these to
tor-consensus-health@ if anybody there wants the notices.

Cheers! -Damian

[1] https://trac.torproject.org/projects/tor/ticket/18177
[2] https://gitweb.torproject.org/doctor.git/tree/fallback_directories.py