Forum Discussion

_V_aniac_74372's avatar
_V_aniac_74372
Icon for Nimbostratus rankNimbostratus
May 13, 2010

Weird Trunk Behavior

I have several pairs of LTM's connecting to Cisco to 2960G48TLL switches running code 12.2.25. The LTM's are on version 10.x. I have configured a trunk on each LTM and I have added interfaces 1.1 and 1.2 to each trunk. I am using tagged VLAN's. I am not using LACP and the interfaces are hard set to 1000/full at both the switch and LTM. The rest of the configurations on the trunk and interfaces are set to the default values.

 

 

The interfaces are connected and reachable via pinging. I have a continuous ping reaching the devices and then after a couple of hours it drops and times out. Then after another long period it re-establishes connectivity and is pingable again. Anyone ever seen this behavior? I have about 5 pairs that keep behaving the same way. My company is not open to using LACP and apparently there are devices configured in this fashion that are working in production. Anyone ever seen this type of behavior? Any ideas? Thanks...below you will see the port-channel config on the switch and the bigip_base.conf.

 

 

Switch Config:

 

 

interface Port-channel1

 

description sf45-iaedev-ltm1

 

switchport access vlan 999

 

switchport trunk allowed vlan 26,27,30,32,36,37,39,117,118

 

switchport mode trunk

 

end

 

!

 

interface GigabitEthernet0/12

 

description sf45-iaedev-ltm1-i1

 

switchport access vlan 999

 

switchport trunk allowed vlan 26,27,30,32,36,37,39,117,118

 

switchport mode trunk

 

speed 1000

 

duplex full

 

no cdp enable

 

channel-group 1 mode on

 

spanning-tree portfast

 

spanning-tree bpduguard enable

 

spanning-tree guard root

 

end

 

!

 

interface GigabitEthernet0/13

 

description sf45-iaedev-ltm1-e1

 

switchport access vlan 999

 

switchport trunk allowed vlan 26,27,30,32,36,37,39,117,118

 

switchport mode trunk

 

speed 1000

 

duplex full

 

no cdp enable

 

channel-group 1 mode on

 

spanning-tree portfast

 

spanning-tree bpduguard enable

 

spanning-tree guard root

 

end

 

 

 

Bigip_base.conf:

 

 

interface Port-channel1

 

description sf45-iaedev-ltm1

 

switchport access vlan 999

 

switchport trunk allowed vlan 26,27,30,32,36,37,39,117,118

 

switchport mode trunk

 

end

 

!

 

interface GigabitEthernet0/12

 

description sf45-iaedev-ltm1-i1

 

switchport access vlan 999

 

switchport trunk allowed vlan 26,27,30,32,36,37,39,117,118

 

switchport mode trunk

 

speed 1000

 

duplex full

 

no cdp enable

 

channel-group 1 mode on

 

spanning-tree portfast

 

spanning-tree bpduguard enable

 

spanning-tree guard root

 

end

 

!

 

interface GigabitEthernet0/13

 

description sf45-iaedev-ltm1-e1

 

switchport access vlan 999

 

switchport trunk allowed vlan 26,27,30,32,36,37,39,117,118

 

switchport mode trunk

 

speed 1000

 

duplex full

 

no cdp enable

 

channel-group 1 mode on

 

spanning-tree portfast

 

spanning-tree bpduguard enable

 

spanning-tree guard root

 

end

 

 

1 Reply

  • Hamish's avatar
    Hamish
    Icon for Cirrocumulus rankCirrocumulus
    Yes, you (probably) have a cabling problem...

     

     

    What do you mean 'your company is not open to using LACP'? If you're using channels with no signalling, then your company deserves a kicking. NEVER EVER configure a channel in production without singalling. It may work today, but it probably wont' work tomorrow. And you'll spend your days trying to discover why. Your company is having a laugh if they think this is going to work well. (And I don't care if its working in production today. Like I said, it probably won't tomorrow... And etherchannel problems without signalling are the worst things to try & debug. Especially when production traffic is on them and your company starts yelling at you to fix it).

     

     

    Sorry...

     

     

    If it helps, you could always point them at the cisco docs that also advise you not to configure channels without signalling...

     

     

    FWIW you get this if you try to configure channels with ESX as well... That does channelling and doesn't understand LACP at all...

     

     

    H