On Thu, Jan 3, 2013 at 5:54 PM, Smith, Donald <[email protected]>wrote: > > >-----Original Message----- > >From: [email protected] [mailto:[email protected]] On Behalf Of > >Tony Li > >Sent: Thursday, January 03, 2013 3:31 PM > >To: Jeff Wheeler > >Cc: [email protected]; [email protected] > >Subject: Re: [GROW] [Idr] I-D Action: draft-ietf-grow-ops-reqs-for-bgp- > >error-handling-06.txt > > > > >I support the general concept of improved error handling and softer > >failures when possible. > > > It exasperated it by causing peers to drop and having to rebuild ribs/fibs > etc... continuously. > > Other than getting it noticed I doubt a reset will ever make a misbehaving > router stop misbehaving.
The "continuously" points to the problem. If a session drops because of syntactic issues, it should stay down. Does that make sense? It is analogous to "treat-as-withdraw", just using a much larger hammer. :-) > >Semantic errors would be consistency violations within the contents of > >the message. In these cases, treat-as-withdraw seems reasonable. > I would also like the ignore option here. You know your neighbor is saying > something you don't understand, possibly due to a lack of in your > vocabulary but you understand other elements so you nod and continue > listening to the rest:) > > This is what happened recently. ALU announced attributes with 1 bit set > that others didn't understand. If the routers that didn't understand that > bit could have ignored just that element everything would have been fine. Here's a couple of problems with what you're suggesting here. You don't know whether that 1 bit changes the logic that should be used (e.g. "This is a back-up route only", or "Black-hole this prefix, please".) Presuming that you should use the announcement is dangerous. And then what should you do with the announcement - forward to appropriate peers, or not? If you forward it, should you keep the bit set, or not? IMHO, the safe thing to do with "unknown bit set", is treat-as-withdraw. Maybe, MAYBE, have a knob to "listen to but ignore unknown bit". DEFINITELY, do NOT announce bits you don't understand. (liberal accept, conservative send). And PROBABLY do NOT announce things if you twiddled unknown bit(s). You are potentially messing with unknown semantics, to the probable detriment of your peers. (Here there be dragons.) The safest thing is to not announce. The conservative approach is to "treat as withdraw". And put a suitable error message wherever error messages normally go. Again, IMHO. Brian
_______________________________________________ GROW mailing list [email protected] https://www.ietf.org/mailman/listinfo/grow
