From: Justin Menga (Justin.Menga@xxxxxxxxxxxxxxxxxx)
Date: Tue Dec 12 2000 - 08:46:58 GMT-3
APOLOGIES - xianqi is correct!!
The config below only worked because it routed ALL PACKETS to the nexthop
specified. I lost adjacencies on ethernet side of the spoke routers - all
multicasts were being channelled to the next hop!!
I dug into the CD doc - match interface xx is only used for route-maps
associated with route redistribution (thus in this scenario it is
ignored....)
Glad I found this out before the lab!!!
So the solution would be to use an access-list only in your route-map (you
can also match on packet length).
-----Original Message-----
From: Justin Menga
Sent: Tuesday, December 12, 2000 11:37 PM
To: 'zhangxianqi'; 'ccielab@groupstudy.com'
Subject: RE: An equivalent to NEXT_HOP_SELF for an IGP?
Hi,
I'm sorry but that just is not true.
Policy route TYPICALLY routes based on source address, but is flexible
enough to match next hop interface.
I had the following in a frame relay NBMA network (10.1.10.x/24):
R6
|
|
R1----R5
I don't have any frame maps - all using InARP. Obviously R6 can't ping R5
and vice versa.
So I added the following to each spoke router:
route-map NEXTHOP permit 10
match interface Serial0
set ip next-hop 10.1.10.1
ip local policy route-map NEXTHOP
After this, R5 and R6 ping each other no problem.
See CCBOOTCAMP Lab 1 for similar scenario, although the solution uses
access-lists - the above I believe is much more elegant but probably
increases processing a little as ALL packets must be policy routed out s0.
If your next hop is your Hub router, the packet does not require policy
routing.
-----Original Message-----
From: zhangxianqi [mailto:zhangxqi@gitc.com.cn]
Sent: Tuesday, December 12, 2000 3:39 PM
To: Justin Menga
Subject: Re: An equivalent to NEXT_HOP_SELF for an IGP?
hi, Menga,
you can write the command at route-map,but policy route will ignore
it,because the policy route according to the source address (if you use
extend access,then both souce and destin address) or packet length to route
packet,so 'match inter face' no use at all.
Regards
xianqi
----- Original Message -----
From: Justin Menga <Justin.Menga@computerland.co.nz>
To: 'zhangxianqi' <zhangxqi@gitc.com.cn>
Sent: Monday, December 11, 2000 6:22 PM
Subject: RE: An equivalent to NEXT_HOP_SELF for an IGP?
> Yes you can......I have tested on many routers...
>
> -----Original Message-----
> From: zhangxianqi [mailto:zhangxqi@gitc.com.cn]
> Sent: Monday, December 11, 2000 2:40 PM
> To: Justin Menga
> Subject: Re: An equivalent to NEXT_HOP_SELF for an IGP?
>
>
> hi Justin,
> You can't use 'match nterface s0' at policy route,you can try it.
>
> Regards
> xianqi
> ----- Original Message -----
> From: Justin Menga <Justin.Menga@computerland.co.nz>
> To: 'zhangxianqi' <xianqizhang@sina.com>; Ronnie Royston
> <RonnieR@globaldatasys.com>
> Cc: <ccielab@groupstudy.com>
> Sent: Sunday, December 10, 2000 6:36 PM
> Subject: RE: An equivalent to NEXT_HOP_SELF for an IGP?
>
>
> > Hi,
> >
> > I think xianqi is talking about policy routing:
> >
> > ip local policy route route-map NEXT_HOP
> >
> > route-map NEXT_HOP permit 10
> > match interface s0
> > set next-hop 10.10.1.4
> > route-map NEXT_HOP permit 20
> >
> > The above basically says if the next hop interface is out s0, then set
the
> > next hop to be 10.10.1.4 (R4). This applies to traffic originating from
> the
> > router....
> >
> >
> >
> > -----Original Message-----
> > From: zhangxianqi [mailto:xianqizhang@sina.com]
> > Sent: Saturday, December 09, 2000 4:35 PM
> > To: Ronnie Royston
> > Cc: ccielab@groupstudy.com
> > Subject: Re: An equivalent to NEXT_HOP_SELF for an IGP?
> >
> >
> > you can do it on redistribut use route-map,in the route-map,set
next-hop.
> > I don't try it on my router,just from memory,sorry if i am wrong.
> > regards
> > xianqi
> > ----- Original Message -----
> > From: Ronnie Royston <RonnieR@globaldatasys.com>
> > To: <ccielab@groupstudy.com>
> > Sent: Saturday, December 09, 2000 8:22 AM
> > Subject: An equivalent to NEXT_HOP_SELF for an IGP?
> >
> >
> > > Is there a way to redistribute routes between IGPs in such a way as to
> set
> > > redistributed routes to point to this router as the next hop?
> > >
> > > I set up a lab where a multipoint subinterface is connecting R7 as
> below:
> > >
> > > R3 R4
> > > \ /
> > > \ /
> > > R7
> > >
> > > R7 shares OSPF area0 with R3 and speaks RIPv2 to R4 (passive interface
> > with
> > > neighbor command). R3 is getting the routes from RIP redistributed
just
> > > fine and as expected, with R4's serial interface as the next hop.
> > However,
> > > routing fails because I only have R3 and R4 frame mapped to R7 and not
> > each
> > > other. I want to use a route-map in redistributing into OSPF to tell
> the
> > > OSPF routers to use R7s serial interface as the next hop for the
routes
> > > learned via RIP. I can't get this to work. How should I resolve
this,
> > or,
> > > is this a completely useless senario and there are no "knobs" to turn
to
> > fix
> > > this? Thanks everyone.
> > >
> > >
> > >
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 08:26:01 GMT-3