(RADIATOR) Port-Error

[Neale Banks]

On Wed, 16 Oct 2002, Mohammed AbdusSami wrote:

> Following is the stop record of my accounting configuration. The cause
> of termination of this is PORT-ERROR. Can you please me what does it
> means.

As Hugh sugested, this is really a NAS-specific issue.  But...

> Mon Oct 14 04:12:16 2002
> 	NAS-IP-Address = 212.26.73.240
> 	NAS-Port = 132
> 	NAS-Port-Type = Async
> 	User-Name = "zajil120771 at zajil.net"
> 	Called-Station-Id = "3602428"
> 	Calling-Station-Id = "33610711"
> 	Acct-Status-Type = Stop
> 	Acct-Authentic = RADIUS
> 	Service-Type = Framed-User
> 	Acct-Session-Id = "000DDA8A"
> 	Framed-Protocol = PPP
> 	Framed-IP-Address = 212.24.231.64
> 	Acct-Terminate-Cause = Port-Error
> 	Acct-Input-Octets = 980238
> 	Acct-Output-Octets = 9946394
> 	Acct-Input-Packets = 11304
> 	Acct-Output-Packets = 10777
> 	Acct-Session-Time = 8201
> 	Acct-Delay-Time = 0
> 	Timestamp = 1034557936

Is this from a Cisco?  If so, these snippages are probably relevant:

===================================8<===================================
Date: Mon, 2 Apr 2001 08:35:24 -0700
From: Dennis Peng <dpeng at cisco.com>
To: Neale Banks <neale at lowendale.com.au>
Cc: cisco-nas at external.cisco.com
Subject: Re: Acct-Terminate-Cause = Port-Error ?

"Port Error" indicates that there was a failure in PPP keepalives, ie
we were sending out LCP echo requests, and didn't receive a LCP echo
reply response 5 times in a row. Perhaps the lower layer already went
away and this wasn't signalled properly? Or modem was retraining for a
long time? It's hard to say. Is this is on an async interface?
Keepalives are turned off by default on async interfaces, so I assume
you have it enabled?

Dennis
===================================8<===================================


===================================8<===================================
Date: Mon, 02 Apr 2001 16:49:56 -0800 (PST)
From: Aaron Leonard <Aaron at Cisco.COM>
To: Neale Banks <neale at lowendale.com.au>
Cc: Dennis Peng <dpeng at Cisco.COM>, cisco-nas at external.cisco.com
Subject: Re: Acct-Terminate-Cause = Port-Error ?

> On Mon, 2 Apr 2001, Dennis Peng wrote:

> > Neale Banks [neale at lowendale.com.au] wrote:
> > >
> > > Definitely Async, it's even logged by RADIUS as "NAS-Port-Type =
Async".
> > > FWIW, the NAS is configured with virtual profiles enabled and is
part of
> > > a stack.
> >
> > Oh, ok, it is enabled on vaccess interfaces by default. Since you are
> > using virtual-profiles, everyone gets a vaccess interface, which means
> > keepalives are enabled. You can turn it off with the "no keepalive"
> > command on the vtemplate. Of course if you do so, you might mask a
> > lower layer problem. But you might also be working around a buggy
> > client. It's hard to say.

> Assuming that (a) everything is virtualised and (b) keepalives are a
Good
> Thing for ISDN callers then the balance probably falls to leaving
> keepalive enabled?  Alternatively, is there a compromise possible by
> relaxing the keepalive parameters?

Yes, I think it's reasonable enough to relax the keepalive interval.
Say, with an interval of 30s, then the call will drop after 3 minutes
of nonresponsiveness from the peer.  It's very unlikely that a modem
link that hasn't responded for 3 minutes will ever recover.

One downside of keepalives is that they may reset some idle timers.
Hopefully (in current code) we've fixed all the places where IOS
does this, but I believe that Windows will always reset its idle
timer when it gets an LCP ECHOREQ.  So you may be defeating your
clients' idle timer settings (which could be an issue for them if
they're paying per-minute charges.)

> And, yes I also suspect it's a buggy client (possibly a dodgy modem,
> possibly a marginal local-loop 

It's even possible that Windows has hung (perish the thought).

Aaron
===================================8<===================================

Further discussion of this should probably be directed to a NAS-oriented
forum (e.g. http://www.cisco.com/warp/public/471/cisco-nas.html ).

HTH,
Neale.

===
Archive at http://www.open.com.au/archives/radiator/
Announcements on radiator-announce at open.com.au
To unsubscribe, email 'majordomo at open.com.au' with
'unsubscribe radiator' in the body of the message.