Re: [Idr] RFC-4893 handling malformed AS4_PATH attributes

On Dec 15, 2008, at 3:00 PM, Jeffrey Haas wrote:
> At the risk of missing something else obvious in a quick reply,  
> presume
> instead of dropping the update that the withdrawn routes field is  
> processed
> and that all reachability covered by the path attributes is deleted
> instead of being installed or updated.  I believe this would have the
> effect I was originally intending without leaving stale information in
> the system.

Right, I think that was what Cayle also described and anyway was what  
I meant to agree with. :-)  It's ugly and has its own problems but  
doesn't appear to violate the loop free property assuming you can do  
it right.  One of the "own problems" is that it presumes the error  
you're coping with is one which leaves the update in a state where you  
can reliably dig out all the un/reachability information, including  
that in the NLRI field, the WITHDRAWN ROUTES field, the  
MP_{UN}REACH_NLRI attributes, and any other place where un/ 
reachability information may end up residing in the future.  (I think  
this is kind of a restatement of your earlier point which generally  
pertained to slippery slopes and grot.)

> I suspect if the WG put our collective heads together we'd probably
> arrive at some consensus on having more survivable sessions.  It'd
> probably result in a re-spin of the previous INFORM specification,
> perhaps renaming the attribute WTF.

You have just volunteered to craft an appropriate TLA back- 
formation. :-)

--John
_______________________________________________
Idr mailing list
Idr@ietf.org
https://www.ietf.org/mailman/listinfo/idr