[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [tor-bugs] #9321 [Tor]: Load balance right when we have higher guard rotation periods

#9321: Load balance right when we have higher guard rotation periods
-------------------------+-------------------------------------------------
     Reporter:  arma     |      Owner:
         Type:  project  |     Status:  needs_review
     Priority:  normal   |  Milestone:  Tor: 0.2.6.x-final
    Component:  Tor      |    Version:
   Resolution:           |   Keywords:  needs-proposal, tor-auth, tor-
Actual Points:           |  client, 026-triaged-1
       Points:           |  Parent ID:  #11480
-------------------------+-------------------------------------------------

Comment (by nickm):

 Notes on file format:
   * A bad line in the file should probably get ignored, not cause the
 whole file to be unparseable.
   * Errors while parsing the file should probably get reported by line
 number.
   * Right now, applying the file to your vote is O(file_size * log
 (n_elts_in_vote)).  As the file gets bigger and bigger, this will take
 more and more time.  I wonder whether it matters.

 Notes on guardfraction voting:
   * Shouldn't we include GuardFraction in consensus votes for all nodes,
 regardless of whether we think they're a Guard?  After all, other
 authorities might decide to vote on whether the node should be a Guard.
   * Why are guardfraction_percentage and its related flag duplicated in
 routerstatus_t and vote_routerstatus_t?  Remember, vote_routerstatus_t
 contains a routerstatus_t.
   * In routerstatus_parse_guardfraction , I'd be more comfortable if we
 checked the return value of strchr.

 Notes on bw calculation:
   * Maybe guard_get_guardfraction_bandwidth should fill in a structure
 rather than allocating one; it's going to get called a lot.

 I need to go back and look at the bandwidth formulas; I didn't check them
 this time around.

 On XXXs:
   * I don't think floor/ceiling matters.
   * It's okay not to assert for that invariant.
   * I don't know about applying guardfraction to weight_for_dir.  Does
 that mean it would apply to choice of directory guards or not?
   * is_possible_guard is not quite equivalent to is_guard; what did you
 mean there?
   * IMO it's fine to ignore smartlist_choose_node_by_bandwidth for now.


 Notes on tests:
   * test_helpers.h needs an #ifdef guard.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/9321#comment:28>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs