[tor-bugs] #1790 [Tor - Relay]: Count bytes spent on answering directory requests

Tor Bug Tracker & Wiki torproject-admin at torproject.org
Thu Aug 5 08:55:49 UTC 2010


#1790: Count bytes spent on answering directory requests
-------------------------+--------------------------------------------------
 Reporter:  karsten      |       Owner:  karsten     
     Type:  enhancement  |      Status:  needs_review
 Priority:  normal       |   Milestone:              
Component:  Tor - Relay  |     Version:              
 Keywords:               |      Parent:              
-------------------------+--------------------------------------------------

Comment(by karsten):

 Replying to [comment:2 nickm]:
 > The check in connection_bytes_decrement is wrong; CONN_TYPE_DIR is not
 only directory request that we're answering, but also directory requests
 that we're making.

 I added a check for conn->purpose == DIR_PURPOSE_SERVER. Look better?

 > The calculation of "len" in rep_hist_get_bandwidth_lines assumed that
 the longest identifier it cared about was write-history; dirreq-write-
 history is longer.

 Right, I added 7 more bytes per line.

 > Also, are there any tors left that still care about getting "opt" ?  I'm
 pretty sure "opt" is unnecessary now that Tor 0.1.2.5 and earlier are
 dead, so we might as well remove it in a followup patch.

 Yes, it would be good to rid of "opt", but, as you say, not as part of
 this patch.

 See new commit in my dirbytes branch that now also contains a changes
 file. I'm running this patch on my test node and will have some results
 later today. Other than the issues you noted here, does this patch look
 good to be merged? If so, and if my test goes well, I'll raise r<2 to r<4
 (TODO line, noted above) and play the squash and rebase game.

 Thanks for the review!

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/1790#comment:4>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online


More information about the tor-bugs mailing list