[33967] in North American Network Operators' Group
Re: History: lengthy outages
daemon@ATHENA.MIT.EDU (Brett Frankenberger)
Thu Jan 25 16:44:33 2001
Message-Id: <200101252140.PAA30961@rbfux.rbfnet.com>
To: sean@donelan.com (Sean Donelan)
Date: Thu, 25 Jan 2001 15:40:57 -0600 (CST)
From: "Brett Frankenberger" <rbf@rbfnet.com>
Cc: nanog@merit.edu
In-Reply-To: <20010125183142.17262.cpmta@c004.sfo.cp.net> from "Sean Donelan" at Jan 25, 2001 10:31:42 AM
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Errors-To: owner-nanog-outgoing@merit.edu
>
> That was a different one, but I didn't have that data readily available.
> The cisco/bay bug I was thinking of was when Cisco's kept reseting all
> their BGP sessions when they saw a particular announcement from Bay routers.
IIRC (or maybe we're thinking of different outages), it was reversed.
There was an advertisement with a malformed AS_PATH attribute "out
there", Cisco's accepted it and passed it on (still in malformed form),
but Bay's (and some other non-Cisco implementations, such as gated)
correctly issued a Notify message and dropped the session upon receipt
of the malformed advertisement.
I believe there were two such outages (separated by a year or so)
before Cisco fixed the bug.
-- Brett
