[tor-bugs] #2485 [Tor Weather]: Review Tor Weather's use of TorCtl

[Tor Bug Tracker & Wiki]

#2485: Review Tor Weather's use of TorCtl
-------------------------+--------------------------------------------------
 Reporter:  mikeperry    |       Owner:  kaner   
     Type:  task         |      Status:  assigned
 Priority:  normal       |   Milestone:          
Component:  Tor Weather  |     Version:          
 Keywords:               |      Points:  2       
   Parent:               |  
-------------------------+--------------------------------------------------
Changes (by mikeperry):

  * status:  new => assigned
  * owner:  mikeperry => kaner


Comment:

 Ok. I've finished a quick review of weather's use of TorCtl. It is pretty
 minimal in its use for the most part. There are some places where it
 manually parses descriptor and ns batches, but for the most part they seem
 ok. In one place, I noticed the check for "opt hibernating" was an
 re.search without the ^ prefix.

 I also noticed that the bandwidth it was checking was the descriptor
 bandwidth and not the consensus bandwidth
 (connection.get_network_status()). We probably should be using the
 consensus values for things like determining the fastest nodes for
 t-shirts. But for slow node alerts, we may want to stick to using
 descriptor values.. Or maybe still use consensus values for them, too.

 Finally, it looks like weather is heavily relying on descriptor presence.
 For this to work out, it really should set FetchUselessDescriptor 1 in its
 torrc, to ensure that it actually downloads descriptors for hibernating
 and downed nodes.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/2485#comment:1>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online