IBM server connected to 4948 - gives error

Unanswered Question
Dec 16th, 2009
User Badges:

Hi


Products


IBM Server with Ethernet Server Connectivity Module, 6 GB Ethernet ports - Qty 1

Cisco 6509 - Oty 02

Cisco 4948 - Qty 02


Earlier setup:  IBM server was connected to 6509-1 and 6509-2 and both switches were connected over a trunk link.  Everything worked fine without any error.


New setup: 4948-1 connected to both 6509 and 4948-2 is also connected to 6509-2.  This gives an error

%C4K_EBM-4-HOSTFLAPPING: Host 00:21:5E:30:0B:0A in vlan 112 is flapping between port Te1/49 and port Gi1/44


                                    Picture2.jpg


In old setup, server MAC is only learned through 6509-1 port.  No spanning-tree blocked ports were on either of switch.


However in new setup, ports from both 4948 towards 6509-2 are in blocked state.


It is clearly a layer 2 loop.  How can I avoid it? And why it is not occuring when directly conneting server to both 6509 switches.


rgds/bsn

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Jon Marshall Wed, 12/16/2009 - 11:12
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

bsn1980in wrote:


Hi


Products


IBM Server with Ethernet Server Connectivity Module, 6 GB Ethernet ports - Qty 1

Cisco 6509 - Oty 02

Cisco 4948 - Qty 02


Earlier setup:  IBM server was connected to 6509-1 and 6509-2 and both switches were connected over a trunk link.  Everything worked fine without any error.


New setup: 4948-1 connected to both 6509 and 4948-2 is also connected to 6509-2.  This gives an error

%C4K_EBM-4-HOSTFLAPPING: Host 00:21:5E:30:0B:0A in vlan 112 is flapping between port Te1/49 and port Gi1/44


                                   


In old setup, server MAC is only learned through 6509-1 port.  No spanning-tree blocked ports were on either of switch.


However in new setup, ports from both 4948 towards 6509-2 are in blocked state.


It is clearly a layer 2 loop.  How can I avoid it? And why it is not occuring when directly conneting server to both 6509 switches.


rgds/bsn


What is port gi1/44 ie. it is not in your diagram.


Jon

bsn1980in Wed, 12/16/2009 - 11:22
User Badges:

Sorry.


There is change in diagram.  Server is connected to Gig1/44 instead of Gig8/45.


rgds/bsn

Jon Marshall Wed, 12/16/2009 - 11:26
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Are both the NICs on the server active or is one active and one standby ?


How are the NICs configured in terms of IP addressing ?


Jon

bsn1980in Wed, 12/16/2009 - 11:32
User Badges:

Server NIC are connected to balde swithces.  The blade switches are connected to 4948 switches.


Server IP is configured as Virtual IP.

Jon Marshall Wed, 12/16/2009 - 11:49
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Can you post diagram again showing blade switches and which links are blocking and which are forwarding.


Jon

Jon Marshall Wed, 12/16/2009 - 12:05
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Sorry to be a pain but your diagram shows the blade switch forwarding on both ports but if it is then you have a L2 loop in your network. One of the blade server uplinks should be blocking.


I'm assuming the server is in vlan 112 and your diagram is showing the links blocking for vlan 112 ?


Jon

Jon Marshall Wed, 12/16/2009 - 12:06
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Follow up to previous post.


Is the blade interconnect blocking ? ie. in which case you don't have a loop.


Jon

bsn1980in Wed, 12/16/2009 - 12:10
User Badges:

No there is no blocking in old case.  When I saw ARP entries, Server MAC can only be learnt on 6509-1.  And on 6509-2, it was learnt through 6509-1. Only one way learning of MAC, hence no loop / error.




rgds/bsn

Jon Marshall Wed, 12/16/2009 - 12:16
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

bsn1980in wrote:


No there is no blocking in old case.  When I saw ARP entries, Server MAC can only be learnt on 6509-1.  And on 6509-2, it was learnt through 6509-1. Only one way learning of MAC, hence no loop / error.




rgds/bsn


So on the blade chassis you have servers and you have a switch at either end of the chassis ?


There is also internally a switch interconnect between the 2 blade switches. Now i'm guessing this must have been blocked because if it wasn't even in the old case you would have a loop.


On each of the switches can you do a "sh spanning-tree vlan 112" and post.


Also the NICs, they share a virtual IP but do they function in active/standby mode ie. only NIC is forwarding at any one time ?



Jon

bsn1980in Wed, 12/16/2009 - 12:26
User Badges:

4948-1#sh spanning-tree vlan 112

VLAN0112

  Spanning tree enabled protocol ieee

  Root ID    Priority    8192

             Address     0024.98ce.a470

             Cost        2

             Port        49 (TenGigabitEthernet1/49)

             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec

  Bridge ID  Priority    32880  (priority 32768 sys-id-ext 112)

            Address     0027.0d83.0f80

             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec

             Aging Time 300

Interface           Role Sts Cost      Prio.Nbr Type

------------------- ---- --- --------- -------- --------------------------------

Gi1/44              Desg FWD 4         128.44   P2p

Te1/48              Root FWD 2         128.49   P2p

Te1/50              Altn BLK 2         128.50   P2p

##########################################################################

4948-2#sh spanning-tree vlan 112

VLAN0112

  Spanning tree enabled protocol ieee

  Root ID    Priority    8192

             Address     0024.98ce.a470

             Cost        2

             Port        49 (TenGigabitEthernet1/49)

             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec

  Bridge ID  Priority    32880  (priority 32768 sys-id-ext 112)

             Address     0027.0d83.12c0

             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec

             Aging Time 300

Interface           Role Sts Cost      Prio.Nbr Type

------------------- ---- --- --------- -------- --------------------------------

Gi1/44              Desg FWD 4         128.44   P2p

Gi1/48              Altn BLK 4         128.48   P2p

Te1/49              Root FWD 2         128.49   P2p

CHN-DC-ACC-SWH-4948-2#

Jon Marshall Wed, 12/16/2009 - 13:18
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Going to need the output off all switches, including the interconnect on the 6500 switches. I suspect the issue may be to do with the introduction of 10Gbps links from the 4948s to the 6500s. What is the speed of the interconnect between the 2 6500 switches ?


Jon

chinkevi_2 Wed, 12/16/2009 - 14:20
User Badges:

can i ask what is the vendor of the switch module in the bladeserver?  If it is an old Dlink i remember it tends to misbehave.

Please also check the spanning tree on the switch module.

if we still don't see the issue, i think we need to dwell into how the bladeserver is configured for internal mngt and external connection.

Actions

This Discussion