[tor-bugs] #13603 [Tor]: too much IOERRORS ?

[Tor Bug Tracker & Wiki]

#13603: too much IOERRORS ?
------------------------+---------------------------
     Reporter:  toralf  |      Owner:
         Type:  defect  |     Status:  new
     Priority:  normal  |  Milestone:
    Component:  Tor     |    Version:  Tor: 0.2.5.10
   Resolution:          |   Keywords:
Actual Points:          |  Parent ID:
       Points:          |
------------------------+---------------------------

Comment (by toralf):

 With the mentioned script I collected the data for my exit node since 31th
 of December. The raw data do contain 12822 fingerprint over the last 25
 days, the current number is 7041 - got with this python script:
 {{{
 #!/usr/bin/python2 -u
 #

 #   Toralf Foerster
 #   Hamburg
 #   Germany

 import sys

 from stem.descriptor.remote import DescriptorDownloader

 def main():
   downloader = DescriptorDownloader()

   try:
     for desc in downloader.get_consensus().run():
       print desc.fingerprint
   except Exception as exc:
     pass

 if __name__ == '__main__':
   main()
 }}}
 Whilst the majority of relay has a reasonable small number of failed
 attempts, there is 1 relay having 1545 counts and a lot more relays having
 a high value of io-errors.
 With the files "all" containing all currently known relays from the python
 script and the raw data from the error collecting script in ioerror.<num>
 I used the following commands for a quick&dirty look at the histogram with
 octave :
 {{{
 $> cat ioerror.* | perl -we 'my %h = (); while (<>) { @f=split();
 $h{$f[9]}++; } @arr = sort { $b->[1] <=> $a->[1] } map { [ $_, $h{$_} ] }
 keys %h; foreach $r (@arr) { @line = @{$r}; print $line[0], "\t",
 $line[1], "\n" }' >io.csv
 $> cat all | while read d; do grep $d io.csv >> io_curr.csv; done
 $> awk ' { print $2 } ' < io_curr.csv > s
 }}}
 and in octave I simply tyüed in :
 {{{
 x=csvread("s");
 hist(x,1000);
 axis([3 200 0 400 ]);
 }}}
 to get the attached histogram.

 BTW : Even If if subtracting the 3000 relays used in the december sybill
 attack there are 2000 relays with a changed nickname in the data set over
 3 weeks.

 If desired I can provide the raw data.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/13603#comment:12>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online