OnionPerf meeting notes

Who: karsten, acute, irl

Ana (acute) and Tom at the University of Aberdeen are working on extending OnionPerf to work with Pluggable Transports as part of an Internews award subgrant. In this session we discussed what changes to OnionPerf would be most useful when handing off this work to the Tor Metrics team.

  • OP should record UNIX timestamps for when large OP downloads have partially completed. These should correspond to sizes of the smaller OP downloads (i.e for a 5m file we should record a timestamp at 50k and 1m). In future this would replace downloading smaller files and decreasing overhead when measuring PTs. #26673
  • check if the 'START' time recorded in the OP Torperf analysis file includes the time used to build a Tor circuit (we currently think this is not the case) #26597
  • allow OP to use both v2 and v3 onion services when performing measurements (applicable to measuring PTs) #29372
  • allow giving OP a key value when building onion addresses. This means we would get the same .onion addresses which could then be monitored easily #29366
  • ensure relays picked first in a circuit have the guard flag #29373
  • OP analysis files sometimes present negative numbers in the payload_progress field - investigate this and fix #29374
  • write extended OP documentation #29375 #29376
  • write extended unit tests for the existing code #29377
  • port OP to Python 3 #29367

Action: irl and acute to file tickets for each bullet point on the Tor trac - DONE

Last modified 2 months ago Last modified on Feb 7, 2019, 9:09:34 AM