Re: [c-nsp] Xconnect on Portchannel interface [EoMPLS]

2009-10-29 Thread Colin Whittaker
On Thu, Oct 29, 2009 at 08:58:17AM +1000, David Hughes wrote:
 This is the configuration we use.  N * GigE port based xconnect on the  
 south side of the PE router and configure the port-channel on the ce  
 device.  Our PEs are multi 10GE attached.  This has worked very well for 
 us.  Only problem is picking up loss of link at the far end.  You really 
 have no choice but to rely on aggressive udld to pick up the circuit loss 
 and take it out of the port-channel.  But, in our case the end systems 
 are 6500s so you can't do fast udld timers.You may have ~~ 1sec 
 reconvergence at L3 but an L2 issue over a pw takes about 20 seconds to 
 be picked up by udld.

Would fast rate LACP help here. Should be able to get lacp to notice
that the link is missing in  5 seconds

Colin
-- 
Colin Whittaker +353 (0)86 8211 965
http://colin.netech.ie  co...@netech.ie
___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


Re: [c-nsp] Xconnect on Portchannel interface [EoMPLS]

2009-10-29 Thread David Hughes


On 29/10/2009, at 5:02 PM, Colin Whittaker wrote:


Would fast rate LACP help here. Should be able to get lacp to notice
that the link is missing in  5 seconds


Interesting thought.  Not that familiar with LACP.  I can say for sure  
that PAgP is relying on IOS's understanding of link state so until the  
interface drops you just black-hole traffic down a port-channel member  
that doesn't go anywhere.  UDLD Aggressive will yank the interface  
down and pull it from the port-channel,   _ eventually _


Does LACP do more end-to-end signalling that could help here?


David
...

___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


Re: [c-nsp] Xconnect on Portchannel interface [EoMPLS]

2009-10-28 Thread David Hughes


On 20/10/2009, at 8:01 PM, Marko Milivojevic wrote:


I had another thought after my original reply, but for some reason I
didn't send you follow-up. Have you tried not enabling EC on Cisco
doing xconnect (PE) at all and simply having it just on end-nodes:

A===PE1---PE2===B



This is the configuration we use.  N * GigE port based xconnect on the  
south side of the PE router and configure the port-channel on the ce  
device.  Our PEs are multi 10GE attached.  This has worked very well  
for us.  Only problem is picking up loss of link at the far end.  You  
really have no choice but to rely on aggressive udld to pick up the  
circuit loss and take it out of the port-channel.  But, in our case  
the end systems are 6500s so you can't do fast udld timers.You may  
have ~~ 1sec reconvergence at L3 but an L2 issue over a pw takes about  
20 seconds to be picked up by udld.



David
...
___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


Re: [c-nsp] Xconnect on Portchannel interface [EoMPLS]

2009-10-20 Thread Lukasz Trabinski

On Fri, 16 Oct 2009, Marko Milivojevic wrote:

Hello.


2009/10/16 Lukasz Trabinski luk...@trabinski.net:

Hello

I have problem with xconnect on portchannel interface.

[...]

How to do xconnect from untagged portchanel interface?


Have you tried disabling LACP on the EC interface and have it
statically configured (mode on)?


Yes, it's works but only from cisco side. On other side of port-chanell we 
have Force10 device and when I configure mode on, I have port-channel UP. 
on cisco device, but down on Force10. The question is, why it's works 
when xconnect is configured on subinterface with tagged vlan? Maybe is it 
conflict with LACP/MPLS signalization?

___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


Re: [c-nsp] Xconnect on Portchannel interface [EoMPLS]

2009-10-20 Thread Marko Milivojevic
 Yes, it's works but only from cisco side. On other side of port-chanell we
 have Force10 device and when I configure mode on, I have port-channel UP. on
 cisco device, but down on Force10. The question is, why it's works when
 xconnect is configured on subinterface with tagged vlan? Maybe is it
 conflict with LACP/MPLS signalization?

I had another thought after my original reply, but for some reason I
didn't send you follow-up. Have you tried not enabling EC on Cisco
doing xconnect (PE) at all and simply having it just on end-nodes:

A===PE1---PE2===B

Enable EC just on A and B and do simple xconnect from all interfaces
on PE1 and PE2? My knowledge is very rusty, but it could be that LACP
will be carried over in port mode.

--
Marko
CCIE #18427 (SP)
My network blog: http://cisco.markom.info/
___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/


[c-nsp] Xconnect on Portchannel interface [EoMPLS]

2009-10-16 Thread Lukasz Trabinski

Hello

I have problem with xconnect on portchannel interface.

Portchannel is up and works correctly when configuration is:

interface Port-channel10
 mtu 9216
 no ip address
 load-interval 30
 storm-control broadcast level 2.00
 storm-control multicast level 2.00
end

interface Port-channel10.508
 no ip address
 encapsulation dot1Q 508
 xconnect 10.50.0.3 508 encapsulation mpls
end


waw-sw1#show mpls l2transport vc 508

Local intf Local circuit  Dest addressVC ID 
Status
-  -- --- -- 
--

Po10.508   Eth VLAN 508   10.50.0.3   508UP

waw-sw1#show interfaces description | inc Po10
Po10   up up 
Po10.508   up up


but I want to do with  untagged vlan:

waw-sw1#show running-config interface port-channel 10
Building configuration...

Current configuration : 211 bytes
!
interface Port-channel10
 mtu 9216
 no ip address
 load-interval 30
 storm-control broadcast level 2.00
 storm-control multicast level 2.00
 xconnect 10.50.0.3 508 encapsulation mpls
end

waw-sw1#show mpls l2transport vc 508

Local intf Local circuit  Dest addressVC ID 
Status
-  -- --- -- 
--

Po10   Ethernet   10.50.0.3   508UP


It's works only 2-3 minutes and after it portchannel change do state down.


Oct 16 15:07:00.404: %LINK-3-UPDOWN: Interface Port-channel10, changed 
state to down
Oct 16 15:07:00.404: %XCONNECT-5-PW_STATUS: MPLS peer 10.50.0.3 vcid 508, 
VC DOWN, VC state DOWN
Oct 16 15:07:02.599: %EC-SP-5-L3DONTBNDL2: Gi1/9 suspended: LACP currently 
not enabled on the remote port.
Oct 16 15:07:08.500: %EC-SP-5-L3DONTBNDL2: Gi1/10 suspended: LACP 
currently not enabled on the remote port.
Oct 16 15:07:11.440: %EC-SP-5-L3DONTBNDL2: Gi1/9 suspended: LACP currently 
not enabled on the remote port.




How to do xconnect from untagged portchanel interface?


--
ŁT___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

Re: [c-nsp] Xconnect on Portchannel interface [EoMPLS]

2009-10-16 Thread Marko Milivojevic
2009/10/16 Lukasz Trabinski luk...@trabinski.net:
 Hello

 I have problem with xconnect on portchannel interface.
[...]
 How to do xconnect from untagged portchanel interface?

Have you tried disabling LACP on the EC interface and have it
statically configured (mode on)?

--
Marko
CCIE #18427 (SP)
My network blog: http://cisco.markom.info/
___
cisco-nsp mailing list  cisco-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/