On Tue, 25 Jan 2000, Martin, Christian wrote:
> I believe that sub-interface MTU settings are irrelevant - as Neil suggests,
> they revert to the framer MTU which resides on the main interface. However,
> some non-physical interfaces allow MTU settings, ie Virtual Templates and
> BVIs. It seems that the SAR handleron ATM interfaces needs to carve PDU
> buffers upon initialization, thus, dynamic MTU adjustment on a per-packet
> basis may not be possible. This is ops experince, though - I don't work for
> Cisco....
>
> HTH,
> -chris
>
> -----Original Message-----
> From: Neil J. McRae [mailto:neil@COLT.NET]
> Sent: Tuesday, January 25, 2000 4:14 AM
> To: Rafi Sadowsky
> Cc: Neil J. McRae; Tony Tauber; Steve Meuse; Jim Warner;
> cisco-nsp@puck.nether.net
> Subject: Re: [nsp] CEF stability
>
>
> On Tue, 25 Jan 2000 11:11:23 +0200 (IST)
> Rafi Sadowsky <rafi@meron.openu.ac.il> wrote:
>
> > on an ATM interface its probably better to enlarge the main interfaces MTU
> > to (at least) the MTU on the sub-interface or strange things may happen
>
> We have the main set at 1500 and the sub at 1500 works fine. having
> a different MTU on the main and sub seems to force the subs to use
> the main interface MTU.
so what's the point of setting the sub-int at a larger MTU then the
main-interface ?
better make it "right" from the start if you need a reload to activate it ...
also I dug the following up from an internal techie mailing list
you may find it useful(ops experince too)
-- Date: Sun, 25 Jul 1999 11:07:53 +0300 From: Simon Shickman <simon@CC.HUJI.AC.IL> Subject: Re: peer restarted<<(context added) ATM main interface>> The default value is "MTU 4470". The problem with this is cisco does not write it in the config file. If you chage the MTU on another (sub)interface, there are instances (bug?) when that value is also applied to an interface set to the default 4470. If we can agree on a value other then 4470, it can be forced into the configuration.
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:09 EDT