RE: [nsp] is line protocol enough?

From: Christian Panigl, ACOnet/UniVie (panigl@cc.univie.ac.at)
Date: Mon Nov 22 1999 - 12:30:02 EST


>Date: Mon, 22 Nov 1999 11:19:12 -0500
>From: "Fabien Berger" <berger@ip-plus.net>
>
>2 routers are connected via a serial v.35 (HDLC encapsulation, keepalives are
>set to 10s). on one end of the line the line protocol is down and on the other
>the line protocol stays up (link is up in both ends). a sh/no sh does not help.

    Looks like keepalives are sent and/or received just in one direction.
    Have a look at you interfaces "last input/last output" timestamps (show
    int).

>the problem is that the static routes pointing towards that interface (with
>line protocol up) are not removed. traffic is lost! does this mean that we
>should never trust the keepalives? should we always let a routing protocol
>discover link failures?

    Even _with_ routing protocols you can have problems in such a situation
    (updates/keepalives getting through just in one direction).
    
    Probably PPP is smarter in detecting a semi-permeable link and turning
    it to down on both ends.
    
    Wkr
    CP



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:07 EDT