[c-nsp] Port-Channel Problem

Dan Armstrong dan at beanfield.com
Mon May 14 16:28:53 EDT 2007


As a followup to this problem I posted about earlier - I've observed
some very strange behaviour that might explain why this GEC went stupid
on me for no apparent reason:


I setup a brand new GEC link, with 1 physical interface in the group.
This was brand new, to a new empty switch, so of course there was no
traffic on the link as show here:


metro2.tor-Front[7609]#sh int gi1/13
GigabitEthernet1/13 is up, line protocol is up (connected)
  Hardware is C6k 1000Mb 802.3, address is 0015.f91d.5c86 (bia
0015.f91d.5c86)
  Description: Facing TOR2-04-4-2GE.915.1oe2 [Gi0/16] (GEC2)
  MTU 1500 bytes, BW 1000000 Kbit, DLY 10 usec,
     reliability 255/255, txload 1/255, rxload 1/255
  Encapsulation ARPA, loopback not set
  Full-duplex, 1000Mb/s, media type is LH
  input flow-control is off, output flow-control is desired
  Clock mode is auto
  ARP type: ARPA, ARP Timeout 04:00:00


But when I look at the port-channel, holy hanna!  What the heck?!  The
traffic load is huge....


metro2.tor-Front[7609]#sh int po115
Port-channel115 is up, line protocol is up (connected)
  Hardware is EtherChannel, address is 0015.f91d.5c86 (bia 0015.f91d.5c86)
  Description: Facing TOR2-04-4-2GE.915.1oe2 [Po1]
  MTU 1500 bytes, BW 1000000 Kbit, DLY 10 usec,
     reliability 255/255, txload 196/255, rxload 180/255
  Encapsulation ARPA, loopback not set
  Full-duplex, 1000Mb/s
  input flow-control is off, output flow-control is unsupported
  Members in this channel: Gi1/13
  ARP type: ARPA, ARP Timeout 04:00:0


Has anybody ever seen this before?  This smells like a bug for sure... I
tried it with LACP, PAgP, and without either, just "on".. same behaviour...


interface GigabitEthernet1/13
description Facing TOR2-04-4-2GE.915.1oe2 [Gi0/16] (GEC2)
no ip address
switchport
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 100
switchport mode trunk
channel-group 115 mode on

!
interface Port-channel115
description Facing TOR2-04-4-2GE.915.1oe2 [Po1]
no ip address
switchport
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 100
switchport mode trunk





Dan Armstrong wrote:

>These were just 2 ports on the same blade of a WS-X6724 blade at both 
>sides... nothing at all strange.
>
>I never thought of not using PAgP or LACP - perhaps I should try it.
>
>I am too nervous to bring the GEC back up - both links out of the 
>Etherchannel have been testing fine for days... maybe I should just suck 
>it up and try it to see if it fails again.
>
>
>
>Mike Lydick wrote:
>
>  
>
>>I had a similar issue when trying to turn up port channels that span 
>>across stack 3750. TAC recommends not using PAGP or LACP. Have not 
>>gotten it work since. Is this similar to your scenerio? Any resolution?
>>
>>----- Original Message ----
>>From: Dan Armstrong <dan at beanfield.com>
>>To: "Collins, Richard (SNL US)" <rich.collins at siemens.com>
>>Cc: cisco-nsp at puck.nether.net
>>Sent: Tuesday, May 8, 2007 7:31:17 PM
>>Subject: Re: [c-nsp] Port-Channel Problem
>>
>>I did exactly that, and managed to get it to go into LACP mode.
>>
>>The Etherchannel ran for about 3 hours without a problem, then all of a
>>sudden started losing pings all over the place.  I took one channel out
>>of service, and it was fine.
>>
>>I tested both physical links separately, and they're both perfect.  I'm
>>scared to put them back into the Etherchannel now for fear that they'll
>>fail again.
>>
>>I am using the single fibre SFPs (the GLC-BX-Us and GLC-BX-Ds) for both
>>of these links.
>>
>>Anybody seen an Etherchannel lose it when the two underlying physical
>>links are seemingly perfect on their own?
>>
>>
>>
>>
>>Collins, Richard (SNL US) wrote:
>>    
>>
>>>So I suppose the opposite side was set at the same time to either
>>>channel-group 10 mode [active or passive] for LACP?
>>>
>>>What about additionally setting..
>>>metro2.tor-Front[760(config-if)#channel-protocol lacp
>>>I can't test this myself but saw the configuration option.
>>>
>>>-Rich
>>>
>>>
>>> 
>>>      
>>>
>>>>Date: Sat, 05 May 2007 02:39:04 -0400
>>>>From: Dan Armstrong <dan at beanfield.com>
>>>>Subject: [c-nsp] Port-Channel Problem
>>>>To: cisco-nsp at puck.nether.net
>>>>Message-ID: <463C2688.8060909 at beanfield.com>
>>>>Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>>>>
>>>>Riddle me this.
>>>>
>>>>I have 1 physical link, and a port-channel interface operating in PAgP
>>>>   
>>>>        
>>>>
>>>mode.
>>> 
>>>      
>>>
>>>>interface GigabitEthernet1/21
>>>>no ip address
>>>>switchport
>>>>switchport trunk encapsulation dot1q
>>>>switchport trunk allowed vlan
>>>>   
>>>>        
>>>>
>>>50,80,119,300-304,349,412,420,440,444,446,447
>>> 
>>>      
>>>
>>>>switchport trunk allowed vlan add 449,500,503,616,620,900
>>>>switchport mode trunk
>>>>channel-group 10 mode desirable
>>>>end
>>>>
>>>>interface Port-channel10
>>>>no ip address
>>>>switchport
>>>>switchport trunk encapsulation dot1q
>>>>switchport trunk allowed vlan
>>>>   
>>>>        
>>>>
>>>50,80,119,300-304,349,412,420,440,444,446,447
>>> 
>>>      
>>>
>>>>switchport trunk allowed vlan add 449,500,503,616,620,900
>>>>switchport mode trunk
>>>>
>>>>
>>>>metro2.tor-Front[7609]#sh int po10
>>>>Port-channel10 is up, line protocol is up (connected)
>>>> Hardware is EtherChannel, address is 0015.f91d.5c8e (bia
>>>>   
>>>>        
>>>>
>>>0015.f91d.5c8e)
>>> 
>>>      
>>>
>>>> Description: GEC to metro1.tor-Mowat [Port-channel10]
>>>> MTU 9216 bytes, BW 1000000 Kbit, DLY 10 usec,
>>>>    reliability 255/255, txload 104/255, rxload 202/255
>>>>
>>>>
>>>>Life was good, then:
>>>>
>>>>
>>>>2 problems.  I first tried to change to LACP:
>>>>
>>>>metro2.tor-Front[760(config-if)#channel-group 10 mode ?
>>>> active     Enable LACP unconditionally
>>>> auto       Enable PAgP only if a PAgP device is detected
>>>> desirable  Enable PAgP unconditionally
>>>> on         Enable Etherchannel only
>>>> passive    Enable LACP only if a LACP device is detected
>>>>
>>>>metro2.tor-Front[760(config-if)#channel-group 10 mode active
>>>>
>>>>
>>>>The interface bounced, and went straight back into PAgP mode.....
>>>>
>>>>I tried it several times.  #$S%E$@#$, always back to PAgP.....
>>>>"channel-group 10 mode desirable"
>>>>
>>>>
>>>>Second problem:
>>>>
>>>>I tried a second link anyway, and when I added a second link into the
>>>>PAgP group, the rely on the port-channel interface started dropping
>>>>   
>>>>        
>>>>
>>>like
>>> 
>>>      
>>>
>>>>a stone,  packets were dropping all over the place and even though
>>>>everything seemed to be up, speed, duplex, vlans, configuration
>>>>perfectly matched between the underlying physical interfaces & the
>>>>port-channel interface.... the po interface was a mess.  The new
>>>>physical link on it's own is clean as a whistle when I setup a test
>>>>vlan, or set both sides up as routed interfaces....
>>>>
>>>>Anybody have any light to shed?
>>>>
>>>>   
>>>>        
>>>>
>>>_______________________________________________
>>>cisco-nsp mailing list  cisco-nsp at puck.nether.net
>>>https://puck.nether.net/mailman/listinfo/cisco-nsp
>>>archive at http://puck.nether.net/pipermail/cisco-nsp/
>>> 
>>>      
>>>
>>_______________________________________________
>>cisco-nsp mailing list  cisco-nsp at puck.nether.net
>>https://puck.nether.net/mailman/listinfo/cisco-nsp
>>archive at http://puck.nether.net/pipermail/cisco-nsp/
>>
>>    
>>
>
>_______________________________________________
>cisco-nsp mailing list  cisco-nsp at puck.nether.net
>https://puck.nether.net/mailman/listinfo/cisco-nsp
>archive at http://puck.nether.net/pipermail/cisco-nsp/
>  
>




More information about the cisco-nsp mailing list