curious DNS outage on March 8th ?

Carter Bullard carter at qosient.com
Wed Mar 28 09:47:22 EDT 2012


Gentle people,
In anticipation of Anonymous crushing the internet, I've been sniffing around my archives
looking for evidence of Anonymous practice runs, and did see a significant DNS perturbation
on March 8th here in NYC.  Did anyone capture a curious DNS outage at your site
around 2012/03/08.22:35:00 - 2012/03/08.23:45:00 EST?

I experienced a number of North American DNS servers "going away", where they
just stopped responding, but no other traffic seems to have been affected.   Need some
additional verification before I will move past the hypothesis stage.   Would be interesting
to know if anyone else saw a sudden loss of responsiveness from exterior DNS servers in
this time period.

I noticed that google.com was hammered pretty good, so focusing on google.com
DNS resolution maybe informational.  If you are capturing user data, so that you can
grep for specific DNS queries, try this query (all times here are EST5EDT4):

   ra -t 2012/03/08.22:35-2012/03/08.23:45 -R 2012/03/08 -e google.com - udp and port domain

If you were affected, you would see a significant number of DNS queries that had no
responses (dst pkts eq 0).  A more scientific report can be generated using rahisto.1.
You can calculate the frequency distribution of transaction times for all DNS going
to google.com's CIDR block:

   rahisto -H dur 20:0-0.1 -r 03/08 03/09 - udp and port domain and net 216.239.0.0/18

Here are my results for the entire day of 03/07:
rahisto -H dur 20:0-0.1 -r 2012/03/07 - udp and port domain and net 216.239.0.0/18
 N = 492     mean = 0.029719  stddev = 0.014173  max = 0.096320  min = 0.000000
           median = 0.028108     95% = 0.053595
             mode = 0.000000
 Class     Interval         Freq    Rel.Freq     Cum.Freq    
     1   0.000000e+00          7     1.4228%      1.4228%    
     2   5.000000e-03          0     0.0000%      1.4228%    
     3   1.000000e-02         71    14.4309%     15.8537%    
     4   1.500000e-02         76    15.4472%     31.3008%    
     5   2.000000e-02         21     4.2683%     35.5691%    
     6   2.500000e-02         87    17.6829%     53.2520%    
     7   3.000000e-02        117    23.7805%     77.0325%    
     8   3.500000e-02         18     3.6585%     80.6911%    
     9   4.000000e-02          5     1.0163%     81.7073%    
    10   4.500000e-02          0     0.0000%     81.7073%    
    11   5.000000e-02         74    15.0407%     96.7480%    
    12   5.500000e-02          8     1.6260%     98.3740%    
    13   6.000000e-02          2     0.4065%     98.7805%    
    14   6.500000e-02          2     0.4065%     99.1870%    
    15   7.000000e-02          2     0.4065%     99.5935%    
    16   7.500000e-02          1     0.2033%     99.7967%    
    17   8.000000e-02          0     0.0000%     99.7967%    
    18   8.500000e-02          0     0.0000%     99.7967%    
    19   9.000000e-02          0     0.0000%     99.7967%    
    20   9.500000e-02          1     0.2033%    100.0000%  

Here are my results for the entire day of 03/08:
rahisto -H dur 20:0-0.1 -R 2012/03/08 - udp and port domain and net 216.239.0.0/18
 N = 1046    mean = 0.021668  stddev = 0.019670  max = 0.092698  min = 0.000000
           median = 0.018990     95% = 0.053714
             mode = 0.000000
 Class     Interval         Freq    Rel.Freq     Cum.Freq    
     1   0.000000e+00        355    33.9388%     33.9388%    
     2   5.000000e-03          0     0.0000%     33.9388%    
     3   1.000000e-02         68     6.5010%     40.4398%    
     4   1.500000e-02        103     9.8470%     50.2868%    
     5   2.000000e-02         23     2.1989%     52.4857%    
     6   2.500000e-02        104     9.9426%     62.4283%    
     7   3.000000e-02        199    19.0249%     81.4532%    
     8   3.500000e-02         22     2.1033%     83.5564%    
     9   4.000000e-02          7     0.6692%     84.2256%    
    10   4.500000e-02          9     0.8604%     85.0860%    
    11   5.000000e-02        123    11.7591%     96.8451%    
    12   5.500000e-02         10     0.9560%     97.8011%    
    13   6.000000e-02          7     0.6692%     98.4704%    
    14   6.500000e-02          2     0.1912%     98.6616%    
    15   7.000000e-02          3     0.2868%     98.9484%    
    16   7.500000e-02          4     0.3824%     99.3308%    
    17   8.000000e-02          5     0.4780%     99.8088%    
    18   8.500000e-02          0     0.0000%     99.8088%    
    19   9.000000e-02          2     0.1912%    100.0000%    
    20   9.500000e-02          0     0.0000%    100.0000%     


From a daily report, you see on the 8th a shift of 1.4% failed transactions to 33.9%.

The incident lasted about an hour, and here is my rahisto.1 run during the event.  99.7% failure.

rahisto -H dur 20:0-0.1 -r 2012/03/08 -t 2012/03/08.22:35-2012/03/08.23:45 - \
         udp and port domain and net 216.239.0.0/18
 N = 335     mean = 0.000245  stddev = 0.004471  max = 0.081955  min = 0.000000
           median = 0.000000     95% = 0.000000
             mode = 0.000000
 Class     Interval         Freq    Rel.Freq     Cum.Freq    
     1   0.000000e+00        334    99.7015%     99.7015%    
     2   5.000000e-03          0     0.0000%     99.7015%    
     3   1.000000e-02          0     0.0000%     99.7015%    
     4   1.500000e-02          0     0.0000%     99.7015%    
     5   2.000000e-02          0     0.0000%     99.7015%    
     6   2.500000e-02          0     0.0000%     99.7015%    
     7   3.000000e-02          0     0.0000%     99.7015%    
     8   3.500000e-02          0     0.0000%     99.7015%    
     9   4.000000e-02          0     0.0000%     99.7015%    
    10   4.500000e-02          0     0.0000%     99.7015%    
    11   5.000000e-02          0     0.0000%     99.7015%    
    12   5.500000e-02          0     0.0000%     99.7015%    
    13   6.000000e-02          0     0.0000%     99.7015%    
    14   6.500000e-02          0     0.0000%     99.7015%    
    15   7.000000e-02          0     0.0000%     99.7015%    
    16   7.500000e-02          0     0.0000%     99.7015%    
    17   8.000000e-02          1     0.2985%    100.0000%    
    18   8.500000e-02          0     0.0000%    100.0000%    
    19   9.000000e-02          0     0.0000%    100.0000%    
    20   9.500000e-02          0     0.0000%    100.0000%   

What I saw was most of my external DNS servers went away for about an hour.  Most everything
else worked fine, such as existing long lived flows, were unaffected.   But web pages that were
dependent on DNS resolution to complete were stalled.

Hope all is most excellent,

Carter

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://pairlist1.pair.net/pipermail/argus/attachments/20120328/578fbd04/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4367 bytes
Desc: not available
URL: <https://pairlist1.pair.net/pipermail/argus/attachments/20120328/578fbd04/attachment.bin>


More information about the argus mailing list