At 12:47 PM 12/29/2003, Manuel Kasper wrote:
>OK, you got me worried - I did some testing with a net4511, and sure 
>enough, I was able to kill it with a single TCP connection at maximum 
>speed (which turned out to be about 16 Mbps) after only 25 seconds. It 
>just reset, no kernel panic, no messages, no nothing. Scary! I tried with 
>BIOS versions 1.15a and 1.22 - no difference (as expected).

Sounds like an overrun to me.  That would explain why a net4801 wouldn't 
crash--it's a lot faster.  Is it possible to do a netstat about 10 seconds 
in and see how deep the send queue is?  Also, it would be interesting what 
effect traffic shaping would have.  I suspect limiting the input side to 
something like 12Mbps would fix it.  Does specifying a larger than default 
value for NMBCLUSTERS and NMBUFS change the behavior?  If all this is true, 
it would mean that FreeBSD is not dealing well with a too deep send 
queue.  That would be worth a bug report.  It should at least punt out a 
message before barfing; dropping packets and not crashing would be even better.

I would point out, however, that the lowly 486 in the net4501 still runs 
something like ten times as fast as a T1.  Another sign that the FreeBSD 
network stack rocks.


         -crl
--
Chad R. Larson (CRL22)    chad at eldocomp dot com
   Eldorado Computing, Inc.   602-604-3100
      5353 North 16th Street, Suite 400
        Phoenix, Arizona   85016-3228

-- CONFIDENTIALITY NOTICE --

This message is intended for the sole use of the individual and entity to whom it is addressed, and
may contain information that is privileged, confidential and exempt from disclosure under applicable
law. If you are not the intended addressee, nor authorized to receive for the intended addressee,
you are hereby notified that you may not use, copy, disclose or distribute to anyone the message or
any information contained in the message. If you have received this message in error, please
immediately advise the sender by reply email, and delete the message. Thank you.