I don't want to be assigned the task of assigning blame -- every vendor has
had their turn receiving and giving it -- but it's pretty clear to me that
because Apple implemented RFC 4436 the bug in Cisco's software was brought
to light.  
 
Since there appears to be no bug in the Apple code, but only in Cisco's ARP
handling in specific cases, I presume that's why Cisco is taking the heat.
 
Regards,
 
Frank

  _____  

From: Prof. Robert Mathews (OSIA) [mailto:[EMAIL PROTECTED] 
Sent: Monday, August 06, 2007 1:24 PM
To: [email protected]
Subject: Re: [WIRELESS-LAN] iPhones, ARP Storms and Network Interruptions
...


Frank Bulk - iNAME wrote: 

Robert:



I'm not exactly sure what you're asking in this first question, but let me
say that not many wireless clients (that would be network stack specific, I
believe) have implemented DNAv4.  That Cisco didn't catch this bug during
their testing process is unfortunate of course, but not altogether abnormal.
Programmers write code all the time that works under normal circumstances,
but when something new is introduced, in a way that the programmer didn't
expect,  bad things can happen, and Cisco is not immune to that.  Aruba's
opportunistic release this week made it clear that they don't suffer from
that bug. 



Dear Frank:

First, I apologize for the tardiness in my response to you, and the list.  I
have been otherwise indisposed, where access to my E.mail was simply not
possible.   The point that I was trying to make WRT my first question was:
it has come to my attention that Apple has had a running problem with DHCP
and DNAv4 implementations.  The thread at:
http://lists.sans.org/pipermail/unisog/2007-January/027056.html  was
recently brought to my attention by someone who is familiar with the issue.
>From the post, it appears that the issue has been longstanding.  

What I am having a problem reconciling is that the pronouncements have been
that the problem was attributable wholly to CISCO.  It appears (at least by
the aforementioned URL) that Apple has had a contributory issue.  IF this is
indeed the case, then the fact that CISCO has issued a patch to combat ARP
storms is only likely to satisfy how CISCO's hardware is able to negotiate
Apple's issue, and as described. 

Yes, bad things can/do happen but, it can happen from MULTIPLE directions.
I am just trying to understanding if CISCO code, and CISCO code alone is at
the bottom of this.





As for your second point, no, a single device is unlikely able to generate
11,000 ARP requests/second.  What was happening was that the ARP traffic was
exiting one WLC onto the wired network, entering another WLC on the same L2
network, and because of the bug, if the client's state had not aged out, was
being spit out again on the wired network, after which the packets entered
the first controller from the wired network and because of the bug, spit it
out again, rinse and repeat.  There's no TTL with ARP and the packets not
crossing any L3 boundaries, even this packet did have one.  So the traffic
would keep cycling around until the wireless client's state expired on the
second (and third, etc) WLC.



Frank

Understood..

All my best,
Robert.
--



**********
Participation and subscription information for this EDUCAUSE Constituent Group 
discussion list can be found at http://www.educause.edu/groups/.

Reply via email to