The MPLS WG Archive

[Adrian Farrel <AF@dataconnection.com>]

Hi,

Just to clarify what Bill has said...


>-----Original Message-----
>From: Sanford, Bill [mailto:bills@netplane.com]
>Sent: Tuesday, November 14, 2000 4:32 PM
>To: 'Rita Hui'; mpls@UU.NET
>Subject: RE: Loose ERO subobject
>
>
>
>
>-----Original Message-----
>From: Rita Hui [mailto:huil_98@yahoo.com]
>Sent: Monday, November 13, 2000 9:21 PM
>To: mpls@UU.NET
>Subject: Loose ERO subobject
>
>
>Hi,
>
>In RSVP-TE, when a LSR receives a PATH message with
>the first ERO subobject being a LOOSE one, which of
>the following behaviour is correct?
>
>1) It generates error "Bad Initial subobject" if the
>node does not belong to the address described by the
>subobject.
>2)  It does not generate the "bad initial subobject"
>error, but directly send the message off towards that
>address.  If it does not have a path, generate "Bad
>loose node".
>
>*****
>Rita, if the LSR gets the PATH message and strips off its own 
>hop address, 

This is the key point in Rita's question.  The LSR, when it 
receives a Path with ERO, expects to be a member of the abstract
node described by the top subobject.  This is not a loose/strict
issue, but rather one of specific address versus prefix/AS number.

If, at the previous LSR, the top subobject was loose and the 
current LSR is not part of the abstract node defined by that
subobject, the previous LSR MUST insert a new subobject to
define an abstract node that includes the current LSR (this 
would normally be an explicit address).

>and the next hop it has is loose, the packet 
>should send it out to the next hop (if it is in its table) or
>send it out an interface or interfaces to establish a 
>connection to the next hop.  The LSR *should* add ERO hops if 
>it can find the next ERO at more than the next hop away.  If it
>can't find it, the LSR should return with an error of "No Route
>Found" with say a "bad loose node" error code.
>****
>
>If it does evaluate the first ERO subobject even if it
>is loose, this is different than what CR-LDP, which
>only evalutes when the first ERO subobject is strict. 
>What is the rationale for this distinct behaviour?
>
>*****
>I would think that this would keep dynamic rerouting working 
>at a particular point in the network, if there were known 
>problems at some of the next hops of the LSR.
>*****
>
>Rita

--
Adrian Farrel  mailto:af@dataconnection.com
Network Convergence Group
Data Connection Ltd., Chester, UK
http://www.dataconnection.com/
Tel: +44 (0) 1244 313440  Fax: +44 (0) 1244 312422