[c-nsp] Problems with dot1q trunk over EoMPLS with WS-X6148-GE-TX

Marco Matarazzo marmata at gmail.com
Sun Aug 8 07:02:46 EDT 2010


Hi all,

was trying to configure an EoMPLS link between two 6500s:

Router1
6506 w/VS-S720-10G IOS 12.2(33)SXI2
Customer facing blade: WS-X6148-RJ-45

Router2
6503 w/WS-SUP32-10GE-3B IOS 12.2(33)SXI2
Customer facing blade: WS-X6148-GE-TX

The routers are connected between via the Sup integrated 10Gb interface, mtu
on them is 9000.

EoMPLS works fine if there's no dot1q trunk going over the VC. If there's
one set, everything SEEMS to work, pings go thru, dns requests are fine, I
can access any vlans from anywhere etc. Problems is with that I cannot
access any internet page of downloading anything, all the connections stall!
Seems like a MTU problem to me so begin troubleshooting and find that the
maximum packet size that can travel between this dot1q trunk over EoMPLS is
1496 instead of 1500.

On both routers of course the VC is up:

Router1#sh mpls l2 vc

Local intf     Local circuit              Dest address    VC ID      Status
-------------  -------------------------- --------------- ----------
----------
Fa2/32         Ethernet                   x.y.z.56   71172104   UP

Router2##sh mpls l2 vc

Local intf     Local circuit              Dest address    VC ID      Status
-------------  -------------------------- --------------- ----------
----------
Gi2/3          Ethernet               x.y.z.40   71172104   UP

And the MTU of the VC is 1500:

Router1##sh mpls l2 vc 71172104 det
Local interface: Fa2/32 up, line protocol up, Ethernet up
  Destination address: x.y.z.56, VC ID: 71172104, VC status: up
    Output interface: Te5/5, imposed label stack {700}
    Preferred path: not configured
    Default path: active
    Next hop: x.y.z.14
  Create time: 03:32:35, last status change time: 03:32:35
  Signaling protocol: LDP, peer x.y.z.56:0 up
    Targeted Hello: x.y.z.40(LDP Id) -> x.y.z.56
    MPLS VC labels: local 969, remote 700
    Group ID: local 0, remote 0
    MTU: local 1500, remote 1500
    Remote interface description: -VC-71172104--
  Sequencing: receive disabled, send disabled
  VC statistics:
    packet totals: receive 1959291, send 3518574
    byte totals:   receive 1809500293, send 700321865
    packet drops:  receive 0, send 0

Router2##sh mpls l2 vc 71172104 det
Local interface: Gi2/3 up, line protocol up, Ethernet up
  Destination address: x.y.z.40, VC ID: 71172104, VC status: up
    Output interface: Te1/2, imposed label stack {969}
    Preferred path: not configured
    Default path: active
    Next hop: x.y.z.13
  Create time: 3d19h, last status change time: 03:30:59
  Signaling protocol: LDP, peer x.y.232.40:0 up
    Targeted Hello: x.y.z.56(LDP Id) -> x.y.z.40
    MPLS VC labels: local 700, remote 969
    Group ID: local 0, remote 0
    MTU: local 1500, remote 1500
    Remote interface description: -VC-71172104--
  Sequencing: receive disabled, send disabled
  VC statistics:
    packet totals: receive 50349195, send 5715589
    byte totals:   receive 10440236044, send 5129079765
    packet drops:  receive 0, send 0

This is the port config:

Router1#sh run int fa 2/32
Building configuration...

Current configuration : 245 bytes
!
interface FastEthernet2/32
 description -VC-71172104--
 no ip address
 ip verify unicast source reachable-via any allow-default
 no ip redirects
 no ip proxy-arp
 xconnect x.y.z.56 71172104 encapsulation mpls

Router2##sh run int gi 2/3
Building configuration...

Current configuration : 257 bytes
!
interface GigabitEthernet2/3
 description -VC-71172104--
 no ip address
 ip verify unicast source reachable-via any
 no ip redirects
 no ip proxy-arp
 speed 100
 duplex full
 xconnect x.y.z.40 71172104 encapsulation mpls


Unfortunately I cannot bump up the mtu on WS-X6148-GE-TX (need the A version
for that!), but this is the port where the xconnect is terminating, so I was
under the impression that I wouldn't need jumbo frames support as the labels
would just be  passed thru the TenG mpls enabled interfaces, isn't it? I
verified that lowering the interface mtu of the client machines makes
everything work again. Played with the mpls mtu command, but it does not
seem to have any effect whatsoever.
Oddly enough, I see giants increasing on Router1, but not on Router2. I
assume these are the dot1q trunk packets, but then why I'm seeing the
counter increasing only on one side? The customer says on his switch
interfaces, the mtu is 1500 on both trunks.
So do you think I really need to bump the blade to at least WS-X6148A-GE-TX
for this config to work, or am I missing something else?

Thanks!
]\/[arco
-- 
I'm Winston Wolf, I solve problems.


More information about the cisco-nsp mailing list