IEOC - Internetwork Expert's Online Community

I have a problem with the LDP on R1 and R2 flapping in the CSC labs (advanced tech).  Has anyone else noticed this.  This is the lab where you peer from the AS1 vrf on router R2 to the global on R1 via OSPF.

I did a debug all and I see some bad tcp window sizes occasionally, also I see that I am sending the keepalives, but I don't see responses on R2.

These are also the same labs where I see path unknown in the mpls forwarding-table on the 7200.

I am using gns3 and the recommended 7200 IOS from the blueprint.

I have tried: bouncing the routers, clearing BGP, clearing ldp neighbor, shut no/shut the interface.  What's crazy is that the end to end  connectivity works as long as the LDP isn't broken while I am pinging.  I am thinking about upgrading the code to something newer in the same 12.2S train for the 7200.

Can't remember what the models answer told, but have you tried to use the connected interface as a source for neighbor establishment?

Router 2 uses the connected interface because that's the only one in the VRF.  Router 1 is using the loopback.  I can specify the connected interface on R1 and see what happens.

Try that. Then you rule out possible IGP flapping, atleast.

Well the forwarding table is acting right, but the ldp is still acting whack.

Rack1R2#sh mpls ldp neighbor vrf AS_1
    Peer LDP Ident: 150.1.1.1:0; Local LDP Ident 150.1.12.2:0
        TCP connection: 150.1.1.1.646 - 150.1.12.2.11016
        State: Oper; Msgs sent/rcvd: 13/2; Downstream
        Up time: 00:02:00
        LDP discovery sources:
          Serial2/0.1, Src IP addr: 150.1.12.1
Rack1R2#
00:26:13: %LDP-5-NBRCHG: LDP Neighbor 150.1.1.1:0 is DOWN
Rack1R2#
00:26:19: %LDP-5-NBRCHG: LDP Neighbor 150.1.1.1:0 is UP

Set the LDP RID:

Rack1R1#sh run int s2/0.1
Building configuration...

Current configuration : 125 bytes
!
interface Serial2/0.1 point-to-point
 ip address 150.1.12.1 255.255.255.0
 mpls ip
 frame-relay interface-dlci 102  
end

Rack1R1#sh run | inc mpls ldp
mpls ldp router-id Serial2/0.1

Set the OSFP RID:

Rack1R1#sh run | inc mpls ldp
mpls ldp router-id Serial2/0.1
Rack1R1#sh run | be router o
router ospf 1
 router-id 150.1.12.1
 log-adjacency-changes
 network 150.1.0.0 0.0.255.255 area 0

Debug mpls packets. I remember seeing this in one of the INE labs where R2 is linked to R6 and R1 to R9. What happened was R6 would attempt to communicate with R2 in a label encapsulated packets. For some reason R6 thought it had to push the label even though it was the connected interface.

The solution was to filter the label for the connected subnet. R6 would not install a label for the connected subnet from R2, and therefore R6 would not encapsulate the packet and the session remained up.

weaverap,

I had a similar problem with a 7200 performing as CSC-PE - found some doc on Cisco's Web site that indicated the need for a loopback interface in the vrf to use as the ldp router-id for the vrf. Just searched again briefly, but was not able to find the doc. I had previously tried to use the connected interface for the router-id.

Adding the loopback in the vrf seemed to stabilize the CSC-PE to CSC-CE ldp sessions. Didn't have the same problem with R4/36xx performing the same role.

This was using Graded Labs equipment.

weaverap:

Well the forwarding table is acting right, but the ldp is still acting whack.

Rack1R2#sh mpls ldp neighbor vrf AS_1
    Peer LDP Ident: 150.1.1.1:0; Local LDP Ident 150.1.12.2:0
        TCP connection: 150.1.1.1.646 - 150.1.12.2.11016
        State: Oper; Msgs sent/rcvd: 13/2; Downstream
        Up time: 00:02:00
        LDP discovery sources:
          Serial2/0.1, Src IP addr: 150.1.12.1
Rack1R2#
00:26:13: %LDP-5-NBRCHG: LDP Neighbor 150.1.1.1:0 is DOWN
Rack1R2#
00:26:19: %LDP-5-NBRCHG: LDP Neighbor 150.1.1.1:0 is UP

Well that makes more sense I guess. I couldn't get the LDP to take the serial interface, as you can see.  Also I was doing some i/o all debug on the LDP I saw that R1 wasn't sending keep alives to router 2.  I deduced (big word of the day) that R2 was tearing down the session, but had no problem reestablishing the conn.

I have moved on though.  Too much time in the munush not enough time progressing through labs (I find myself escaping to the board again )  I will probably digress back to the lab some other time and beat it into submission.

What’s weird is my constant forwarding table problem disappeared.  Probably the order that things where brought up?

I couldn't help myself.  I added the loopback to the AS_1 VRF (which apparently populates the ID portion of the TLV, which is probably the buggy part of this code, it doesn't know who to default to the LDP interface).  It appears that is all it takes, although you must then advertise the loopback into OSPF for the LDP adjacency to come up.  I also hard to stop and start the routers, I couldn't figure out a way to clear the neighbors. 
Another interesting thing that I saw is that the lower ip address always sourced/received from the 646 port. 
Now I can sleep like a baby ;)

Rack1R1#sh mpls ldp neighbor
    Peer LDP Ident: 2.2.2.2:0; Local LDP Ident 150.1.1.1:0
        TCP connection: 2.2.2.2.646 - 150.1.1.1.11040
        State: Oper; Msgs sent/rcvd: 15/15; Downstream
        Up time: 00:03:04
        LDP discovery sources:
          Serial2/0.1, Src IP addr: 150.1.12.2
        Addresses bound to peer LDP Ident:
          2.2.2.2         150.1.12.2

Rack1R1#sh mpls ldp neighbor
    Peer LDP Ident: 222.222.222.222:0; Local LDP Ident 150.1.1.1:0
        TCP connection: 222.222.222.222.11006 - 150.1.1.1.646
        State: Oper; Msgs sent/rcvd: 13/13; Downstream
        Up time: 00:00:31
        LDP discovery sources:
          Serial2/0.1, Src IP addr: 150.1.12.2
        Addresses bound to peer LDP Ident:
          150.1.12.2      222.222.222.222

Rack1R2#sh mpls ldp neighbor vrf AS_1
    Peer LDP Ident: 150.1.1.1:0; Local LDP Ident 2.2.2.2:0
        TCP connection: 150.1.1.1.11040 - 2.2.2.2.646
        State: Oper; Msgs sent/rcvd: 21/21; Downstream
        Up time: 00:08:25
        LDP discovery sources:
          Serial2/0.1, Src IP addr: 150.1.12.1
        Addresses bound to peer LDP Ident:
          150.1.1.1       150.1.13.1      150.1.12.1

Rack1R2#sh mpls ldp neighbor vrf AS_1
    Peer LDP Ident: 150.1.1.1:0; Local LDP Ident 222.222.222.222:0
        TCP connection: 150.1.1.1.646 - 222.222.222.222.11006
        State: Oper; Msgs sent/rcvd: 13/13; Downstream
        Up time: 00:00:18
        LDP discovery sources:
          Serial2/0.1, Src IP addr: 150.1.12.1
        Addresses bound to peer LDP Ident:
          150.1.1.1       150.1.13.1      150.1.12.1

links to some other discussion on this.

https://supportforums.cisco.com/message/973022;jsessionid=3A760710A734A6D29E94A5D9DC6C8115.node0?tstart=1

http://www.mail-archive.com/ccie_sp@onlinestudylist.com/msg00378.html