[26788] in North American Network Operators' Group
Re: CEF Load balancing...
daemon@ATHENA.MIT.EDU (Mark Prior)
Sun Jan 16 17:27:43 2000
Message-ID: <200001162224.IAA18151@kuji.off.connect.com.au>
To: Jesper Skriver <jesper@skriver.dk>
Cc: "Alex P. Rudnev" <alex@virgin.relcom.eu.net>,
William Allen Simpson <wsimpson@greendragon.com>, nanog@merit.edu
In-reply-to: Your message of "Sun, 16 Jan 2000 14:51:50 BST."
<20000116145150.A63271@skriver.dk>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <18149.948061478.1@connect.com.au>
Date: Mon, 17 Jan 2000 08:54:38 +1030
From: Mark Prior <mrp@connect.com.au>
Errors-To: owner-nanog-outgoing@merit.edu
> It's almost safe to use it, if you can turn off features which cause troubles.
> CEF is not trouble-making feature except some configurations.
>
> Except when the router decides to turn CEF off all by itself due to
> malloc failure.
That sounds like a memory leak, have you tried the latest versions ? in
12.0(x)S there was a catastrophic memory leak in 12.0(7)S related to cef
...
It's more bizarre than that since show memory indicates its got plenty
so its either recovering it somehow or confused. I would suggest that
later.
We're running the latest code, well not quite (only 12.0(8)S rather
than 12.0(8.3)S, but we are sick of catastrophic failures and 12.0(8)S
seems stable enough, or more exactly we don't need more bug
introducing features.
Mark.
