[VoiceOps] Specific SIP packets cause some Intel gigabit ethernet controllers to reset

Jay Ashworth jra at baylink.com
Wed Feb 6 15:55:02 EST 2013


----- Original Message -----
> From: "Kristian Kielhofner" <kris at kriskinc.com>

> I'm sure some of you have seen this elsewhere but it seems the more
> places I share it the more confirmations I receive:
> 
> http://blog.krisk.org/2013/02/packets-of-death.html
> 
> Long story short some oddly configured Yealink devices could cause our
> Intel gigabit controllers to lose link.

Over here on the voice side of things, this reminds me of a floor-killer
bug I had to chase down, with some help from Mike Cargile, when I was
at VICI Marketing.

We'd upgraded our Asterisk installations from 1.2.24 or 6 to 1.2.30.2, and
everything was fine.

Then we upgraded to 30.4, and progressively, my entire floor would collapse,
at least once a day.

The problem, it turned out after 6 hours of groveling with wireshark, was
that something in the IAX driver was supposed to be resetting a 16-bit
counter, and wasn't (or the reverse; it's been 4 years), and the driver
would get clogged up as old sessions weren't reaped, eventually causing the
entire 255 seat fronter/closer call center to come to a halt.

It took us over a week to finally nail it down; it was intermittent as well,
based on I no longer remember what; some days, we'd be fine.  Some days, we'd
come down 2 or 3 times.

The debugging sagas certainly make great reading, though; thanks again.

Cheers,
-- jra
-- 
Jay R. Ashworth                  Baylink                       jra at baylink.com
Designer                     The Things I Think                       RFC 2100
Ashworth & Associates     http://baylink.pitas.com         2000 Land Rover DII
St Petersburg FL USA               #natog                      +1 727 647 1274


More information about the VoiceOps mailing list