Changes between Version 2 and Version 3 of Ticket #13603, comment 12


Ignore:
Timestamp:
Jan 25, 2015, 3:50:11 PM (5 years ago)
Author:
toralf
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #13603, comment 12

    v2 v3  
    1 With the mentioned script seen above I do collect data for my exit node since 31th of December. The raw data do contain about 650000 lines of about 12800 unique fingerprint over the last 25 days. The current amount of known fingerprints by directory authorities is about 7000 (derived using https://stem.torproject.org/tutorials/mirror_mirror_on_the_wall.html#where-can-i-get-the-current-descriptors )
     1With the mentioned script seen above I do collect data for my exit node since 31th of December. The raw data do contain about 650000 lines of about 12800 fingerprints over the 25 days. The current amount of known fingerprints by directory authorities is about 7000 (derived using https://stem.torproject.org/tutorials/mirror_mirror_on_the_wall.html#where-can-i-get-the-current-descriptors )
    22
    3 Whilst the majority of relay has a reasonable small number of failed attempts, there is 1 relay having 1545 ioerrors.
     3Whilst the majority of relays have a reasonable small number of failed attempts, there is 1 relay having 1545 ioerrors.
    44
    5 With the files "all" containing the nicknames of known relays and the raw data from the error collecting script (file names: ioerror.<num>) I used the following commands for a quick&dirty look at the histogram with gnuplot eventually:
     5With the files "all" containing the fingerprints of all known relays and the raw data files from the error collecting script (file names: ioerror.<num>) I used the following command sequence to produce the histograms with gnuplot as seen in the attachment section:
    66{{{
    77$> cat ioerror.* | perl -we 'my %h = (); while (<>) { @f=split(); $h{$f[9]}++; } while ((my $k, $v) = each %h) { print $k, "\t", $v, "\n" }' | sort > io.csv
     
    1212}}}
    1313
    14 FWIW : There are about 5000 nicknames in the data set over 3 weeks which are now known by the directory authorities.
     14At a first glance the histogram looks like a superposition of 2 different processes. One looks like the well known f(x) = exp(-x). But there seems to be a gaussian curve too with the mean x value of about 100.
    1515
    1616If desired I can provide the raw data.
     17
     18FWIW : There are about 5000 fingerprints in the data set which are not known by the directory authorities. Those were excluded from the dataset (although it wouldn't make any difference).
     19