[tor-bugs] #7359 [Tor]: Design/implement method for collecting/reporting statistics

[Tor Bug Tracker & Wiki]

#7359: Design/implement method for collecting/reporting statistics
------------------------------------------------------------------------+---
 Reporter:  robgjansen                                                  |          Owner:                    
     Type:  enhancement                                                 |         Status:  needs_revision    
 Priority:  normal                                                      |      Milestone:  Tor: 0.2.5.x-final
Component:  Tor                                                         |        Version:                    
 Keywords:  performance, simulation, statistics, tor-relay, tor-client  |         Parent:  #7357             
   Points:                                                              |   Actualpoints:                    
------------------------------------------------------------------------+---

Comment(by robgjansen):

 Replying to [comment:16 nickm]:
 > So, my questions about efficiency and performance are really questions,
 and not rhetorical.  I'm asking: in practice, when you turn this feature
 on and run a reasonably busy simulation, is the performance/memory hit
 acceptable or not?
 >
 > I'm asking that because I suspect that it's going to consume a whole lot
 of memory.  And if it does, we can make it more efficient.  But if it
 doesn't (and you've tested it out, so I'm hoping you will know), then
 there is not any point in putting the effort into making it more
 efficient.
 >

 Of interest here is recent functionality I've added to Shadow. It now has
 the ability to track memory usage of simulated nodes over time (see
 [https://github.com/shadow/shadow/issues/144 this issue on GitHub]). In
 the XML file, you include the 'heartbeatloginfo' attribute like

 {{{
 <node id="relay" heartbeatloginfo="node,ram,socket" ...>
 ...
 }}}

 For this ticket, you probably only need heartbeatloginfo="ram", and only
 on the nodes that you want to profile because Shadow will consume
 significantly more RAM to track the malloced pointers and sizes.

 Once your simulation is complete, the contrib/analyze.py script will parse
 allocated, deallocated, and running total RAM usage over time from the
 scallion.log file, and can also be used to graph the results.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/7359#comment:17>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online