Re: excessive TCP duplicate acks?

From: Andre Guibert de Bruet <andy_at_siliconlandmark.com>
Date: Fri, 23 Feb 2007 02:17:15 -0500
On Feb 21, 2007, at 10:39 AM, Andre Oppermann wrote:

> Andre Guibert de Bruet wrote:
>> On Feb 20, 2007, at 2:31 PM, Andrew Gallatin wrote:
>>> Thanks for reporting this.  I'm glad I'm not the only one who is   
>>> seeing this!
>> I discovered this while testing out a new multi-homed bonded-T1   
>> installation. I was wondering why I was getting no more than  
>> ~170KB/ s. So far, here are the data points that I have gathered:
>> - It affects more than one type of NIC (em0).
>> - It happens regardless of hardware checksum enabled status.
>> - It happens independent of IP aliasing.
>> - I tested GENERIC as of Feb 19 and it is affected.
>> - A kernel from December 15th, 2006 exhibits this problem.
>> - The affected system is connected to a Linksys 24-port unmanaged   
>> gigabit switch (I have the exact same hardware working just fine  
>> with  5.5-STABLE on the same switch).
>
> Today I finally got some spare time and tried to track this down.   
> However
> there is no obvious smoking gun to be seen.  I'll dig further.

Andre,

Just as a followup to my off-list email: I have since come across  
kern/102653. Though my setup uses stateful inspection rules  
exclusively, the problem still persists with "allow ip from any to  
any" as the single rule in ipfw. I have a pcap of the traffic  
patterns that I am seeing, at http://siliconlandmark.com/staff/andre/ 
files/7-current_dupacks.pcap (2.3MB - This pcap was made with the  
single allow any to any rule).

Andy

/*  Andre Guibert de Bruet  * 6f43 6564 7020 656f 2e74 4220 7469 6a20 */
/*   Code poet / Sysadmin   * 636f 656b 2e79 5320 7379 6461 696d 2e6e */
/*   GSM: +1 734 846 8758   * 5520 494e 2058 6c73 7565 6874 002e 0000 */
/* WWW: siliconlandmark.com * C/C++, Java, Perl, PHP, SQL, XHTML, XML */
Received on Fri Feb 23 2007 - 06:17:34 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:05 UTC