From: Carlos G Mendioroz (tron@xxxxxxxxxxx)
Date: Wed May 08 2002 - 08:08:10 GMT-3
One more thing...
I remembered now that in a route reflector lab (hub and 2 spokes), I
noticed that
an intermediate RR (hub) does filter the remote spoke advertisements
while the RFC
says that it should not change the peers (spokes in this case)
communication in any way.
I almost opened a TAC case for that ! :-)
And it turns out that this behaviour prevents the race state...
Carlos G Mendioroz wrote:
>
> Neat. Have reproduced it here too.
> One way out is "preferring" R2 routes over R3 ones (e.g. via weight)
> at R1 and the same thing (simetrical, that is) at R3.
> That makes things stable.
>
> Brian McGahan wrote:
> >
> > Jason,
> >
> > The config is fine, but the BGP session is in race state. This
> > is exactly how it is supposed to act. If you 'debub ip routing' on R1
> > you get:
> >
> > Router1#ping 165.40.23.1
> >
> > Type escape sequence to abort.
> > Sending 5, 100-byte ICMP Echos to 165.40.23.1, timeout is 2 seconds:
> > !!!!!
> > Success rate is 100 percent (5/5), round-trip min/avg/max = 28/30/32 ms
> > Router1#
> > 1w0d: RT: Nexthop for 165.40.22.0/23 updated
> > 1w0d: RT: recursion error routing 165.40.23.1 - probable routing loop
> >
> > Router1#ping 165.40.23.1
> >
> > Type escape sequence to abort.
> > Sending 5, 100-byte ICMP Echos to 165.40.23.1, timeout is 2 seconds:
> > .....
> > Success rate is 0 percent (0/5)
> > 1w0d: RT: recursion error routing 165.40.23.1 - probable routing loop
> > 1w0d: RT: Nexthop for 165.40.22.0/23 updated
> > Router1#ping 165.40.23.1
> >
> > Type escape sequence to abort.
> > Sending 5, 100-byte ICMP Echos to 165.40.23.1, timeout is 2 seconds:
> > !!!!!
> >
> > BGP cannot depend on itself for connectivity like this, hence the
> > session is in race state. Take a look at R1's routing table while the
> > session is in race.
> >
> > 165.40.0.0/23 is subnetted, 1 subnets
> > B 165.40.22.0 [200/0] via 165.40.23.1, 00:00:19
> >
> > That's a recursive loop, it's like saying:
> >
> > Ip route 1.1.1.1 255.255.255.255 1.1.1.1
> >
> > This is the reason you need a full *physical* iBGP mesh, or an IGP to
> > provide underlying BGP transport.
> >
> > BTW, the 'next-hop-self' command on R1 isn't doing anything. That's
> > only for modifying the next-hop for EBGP learned routes.
> >
> > HTH
> >
> > Brian McGahan
> > CCIE #8593
> > brian@cyscoexpert.com
> >
> > CyscoExpert Corporation
> > Internetwork Consulting & Training
> > http://www.cyscoexpert.com
> > Voice: 847.674.3392
> > Fax: 847.674.2625
> >
> > -----Original Message-----
> > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> > Jason Wydra
> > Sent: Monday, May 06, 2002 4:37 PM
> > To: Logan, Harold; Carlos G Mendioroz
> > Cc: Groupstudy ccielab list
> > Subject: RE: IBGP HELP??
> >
> > CONFIGS
> > R1
> > interface TokenRing0/0
> > ip address 204.156.20.5 255.255.255.252
> > ring-speed 16
> > !
> > interface BRI0/0
> > no ip address
> > shutdown
> > cdapi buffers regular 0
> > cdapi buffers raw 0
> > cdapi buffers large 0
> > !
> > interface Serial0/1
> > no ip address
> > shutdown
> > !
> > interface FastEthernet1/0
> > ip address 204.156.20.2 255.255.255.252
> > shutdown
> > duplex auto
> > speed auto
> > !
> > router bgp 3127
> > no synchronization
> > bgp log-neighbor-changes
> > network 204.156.20.4 mask 255.255.255.252
> > neighbor 165.40.23.1 remote-as 3127
> > neighbor 204.156.20.6 remote-as 3127
> > neighbor 204.156.20.6 next-hop-self
> >
> > R2
> > interface Serial0/0
> > no ip address
> > shutdown
> > no fair-queue
> > !
> > interface TokenRing0/0
> > ip address 204.156.20.6 255.255.255.252
> > interface BRI0/0
> > !
> > interface BRI0/0
> > no ip address
> > shutdown
> > cdapi buffers regular 0
> > cdapi buffers raw 0
> > cdapi buffers large 0
> > !
> > interface Serial0/1
> > no ip address
> > shutdown
> > !
> > interface FastEthernet1/0
> > ip address 165.40.23.0 255.255.254.0
> > shutdown
> > duplex auto
> > speed auto
> > !
> > router bgp 3127
> > no synchronization
> > bgp log-neighbor-changes
> > network 165.40.22.0 mask 255.255.254.0
> > network 204.156.20.4 mask 255.255.255.252
> > neighbor 165.40.23.1 remote-as 3127
> > neighbor 204.156.20.5 remote-as 3127
> > R3
> > interface Ethernet0/0
> > ip address 165.40.23.1 255.255.254.0
> > shutdown
> > half-duplex
> > !
> > interface Serial0/0
> > no ip address
> > no fair-queue
> > shutdown
> > no fair-queue
> > frame-relay lmi-type cisco
> > !
> > interface BRI0/0
> > ip unnumbered Ethernet0/0
> > encapsulation ppp
> > shutdown
> > timeout absolute 35000 0
> > dialer idle-timeout 2147483
> > dialer map ip 204.156.20.1 5340
> > dialer-group 1
> > isdn switch-type basic-ni
> > isdn spid1 01532001 5320
> > isdn spid2 01532101 5321
> > cdapi buffers regular 0
> > cdapi buffers raw 0
> > cdapi buffers large 0
> > !
> > interface Serial0/1
> > no ip address
> > shutdown
> > !
> > router bgp 3127
> > no synchronization
> > bgp log-neighbor-changes
> > network 165.40.22.0 mask 255.255.254.0
> > neighbor 165.40.23.0 remote-as 3127
> > neighbor 204.156.20.5 remote-as 3127
> > !
> > "Logan, Harold" <loganh@mccfl.edu> wrote: Would you please post your
> > bgp configs? It would make the discussion a bit easier.
> >
> > Hal
> >
> > -----Original Message-----
> > From: Jason Wydra [mailto:jasonwydra@yahoo.com]
> > Sent: Monday, May 06, 2002 4:01 PM
> > To: Carlos G Mendioroz
> > Cc: Groupstudy ccielab list
> > Subject: Re: IBGP HELP??
> >
> > I'm not running an IGP yet R1 learns about R3 via R2. I have a network
> > statement on R2 for both links to R1 and R2. R1 has the route to R3 in
> > its bgp and ip route table. R3 learns nothing from R2 about networks on
> > the R1 side. Need to put a static route in on R3 pointing to R2 as next
> > hop for the TCP connection to come up.
> > Carlos G Mendioroz wrote: Jason,
> > What IGP are you running ?
> > How is R1 supposed to learn how to reach R3 ?
> >
> > Remember that iBGP will never (well almost never *) relay
> > information, so you need some IGP to do that.
> >
> > (Hmm, you said that R1 has learned via BGP the 165.40.22.0/23,
> > so it has to be originated at R2, may be by a network command.
> > Do you have a simmilar network command for the 204.156.20.0/30 ?)
> >
> > (*) The "almost never" relates to setups like route reflectors,
> > where you get a controlled iBGP redistribution.
> >
> > HTH.
> >
> > Jason Wydra wrote:
> > >
> > > I have AS 3127 with 3 routers. R1 is connected to R2 via Token ring.
> > R2 is connected to R3 via Ethernet. R1 and R3 do not have a direct
> > connection. They are attempting to peer through R2. R1 and R2 neighbor
> > states are active. R2 and R3 neighbor state is active. R1 and R3
> > neighbor state will NOT come up. The network on link between R1 and R2
> > is 204.156.20.0/30 and network between R2 and R3 is 165.40.22.0/23. From
> > R1 I CANNOT ping R3. R1 has learned a route from BGP to R3. Looking on
> > R3 it has not learned a route to R1 (From R2). This is obviously why I
> > cannot ping from R1 to R3 and this is also why the BGP peer won't come
> > up. Simply adding a static route on R3 pointing to R1 solves the problem
> > and my BGP peers comes up. My question is why does R2 tell R1 about the
> > 165 network but R2 does not tell R3 about the 204 network? Why do I have
> > to add a static to R3? Please help!!
> > >
> > > Jason Wydra
> > >
> > > CCNP,CCDP
> > >
> > > ---------------------------------
> > > Do You Yahoo!?
> > > Yahoo! Health - your guide to health and wellness
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:58:52 GMT-3