few questions about etherchannel

Answered Question
Nov 29th, 2008
User Badges:
  • Bronze, 100 points or more

Hi everybody!

I have few questions about etherchannel:


1)Consider an etherchannel consists of two ports. If I change the speed or Duplex on one link in etherchannel ,Would Etherchannel break Or Channel protocol such as pagp/lacp modifies the speed or duplex setting for all the links and etherchannel remains intact? How about if we configure the etherchannel manually(pagp/lacp disabled),what would happen if I change the speed/duplex on one link in etherchannel?


2) Do switches index links in etherchannel identically? For example two switches, sw1 and sw2 are connected by 4- link etherchannel. Sw1 indexes the links in etherchannel as

f0/1-------0

f0/2-------1

f0/3-------2

f0/4-------3.

Will sw2 index its links in the etherchannel identically?

thanks a lot!

Correct Answer by Jon Marshall about 8 years 6 months ago

Sarah


Okay things are not as cut and dried as they seem. You inspired me to do some testing so i dug out 2 3550 switches from the garage and after a quick excursion to get crossover cables (you never have them when you need them) i ran the following test -


I connected the 2 switches with an etherchannel on int fa0/10 & int fa0/11. The ports were set to 10Mbps full. I then


1) Changed int fa0/10 on sw1 to run at 100. The etherchannel on both sides marked the port as down.


2) Change int fa0/10 on sw2 to run at 100. The etherchannel on both sides marked the port as suspended.


At no time did changing the anything on fa0/10 effect the settings on fa0/11.


Not sure of any conclusion we can come to yet


1) Maybe there is an order i need to make the changes in

2) The document i posted previously applied to 4500/6500, perhaps it's different with them but unfortunately i don't have either in my garage :-)



Test output

===========


1) All interfaces on both SW1 & SW2 are running 10Mbps full

===========================================================


SW1#sh etherchannel sum

Flags: D - down P - in port-channel

I - stand-alone s - suspended

H - Hot-standby (LACP only)

R - Layer3 S - Layer2

u - unsuitable for bundling

U - in use f - failed to allocate aggregator

d - default port


Number of channel-groups in use: 1

Number of aggregators: 1


Group Port-channel Protocol Ports

------+-------------+-----------+-----------------------------------------------

1 Po1(SU) PAgP Fa0/10(P) Fa0/11(P)


SW1#


2) Changed int fa0/1 on SW1 from 10Mbps full to 100Mbps full

============================================================


SW1#sh etherchannel summ

Flags: D - down P - in port-channel

I - stand-alone s - suspended

H - Hot-standby (LACP only)

R - Layer3 S - Layer2

u - unsuitable for bundling

U - in use f - failed to allocate aggregator

d - default port


Number of channel-groups in use: 1

Number of aggregators: 1


Group Port-channel Protocol Ports

------+-------------+-----------+-----------------------------------

1 Po1(SU) PAgP Fa0/10(D) Fa0/11(P)


SW1#



3) Changed int fa0/10 on SW2 from 10Mbps to 100Mbps full

========================================================


SW1#sh etherchannel summ

Flags: D - down P - in port-channel

I - stand-alone s - suspended

H - Hot-standby (LACP only)

R - Layer3 S - Layer2

u - unsuitable for bundling

U - in use f - failed to allocate aggregator

d - default port


Number of channel-groups in use: 1

Number of aggregators: 1


Group Port-channel Protocol Ports

------+-------------+-----------+-----------------------------------------------

1 Po1(SU) PAgP Fa0/10(s) Fa0/11(P)


==========================


I was intending to run the test you proposed in your last response and i will when i get the chance.


Jon

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (4 ratings)
Loading.
Giuseppe Larosa Sat, 11/29/2008 - 09:01
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Sarah,

first of all an advice:

I would never touch the configuration of a working etherchannel in a live network.


1) I would expect the modified link to be placed out of the bundle but if you are lucky you can cause worse problems. This should happen regardless the type of etherchannel manual, lacp, pagp: it is a basic consistency check that each switch does on its side


2) Sw2 may or may not use the same indexing scheme: each switch is in charge for load balancing decisions when tramsmitting.

So nothing forbides the two switches to use different indexing schemes


Hope to help

Giuseppe


sarahr202 Sat, 11/29/2008 - 12:40
User Badges:
  • Bronze, 100 points or more

Thanks for your reply Giuseppe!

What load balancing has to do with index scheme?

Load balancing method just determine which link should be used for transmission.

In nut shell, if i am correct, each switch could have different indexes for the link.

For example sw1 might assigns a link in etherchannel index=0, while the same link may have index=1 assigned by sw2.

Am i correct ?

thanks a lot!

Jon Marshall Sat, 11/29/2008 - 10:02
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Sarah


Just to elaborate on Giuseppe's point 1. PagP and LACP are protocols designed to automate the creation of etherchannels but they cannot make changes to the switchports configuration ie. they can only work with what they have so they will not modify settings on the fly.


Jon

sarahr202 Sat, 11/29/2008 - 12:34
User Badges:
  • Bronze, 100 points or more

thanks for your reply!

Here is quote from cisco press book ccnp BCMSN by David Hucaby page no 169 chapter 7

"Pagp also dynamically modifies parameters of the etherchannel if one of the bundled port is modified. For example,if the configured vlan,speed or duplex mode of a port in an established bundle is changed,pagp reconfigures that parameter for all ports in the bundle"

Based on replies, it seems above is wrong.


Jon Marshall Sat, 11/29/2008 - 13:33
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Sarah


No, the above would appear to be correct and we are wrong :-). From Cisco best practices document -


Port Aggregation Protocol

PAgP is a management protocol that checks for parameter consistency at either end of the link and assists the channel in adapting to link failure or addition. Note these facts about PAgP:


PAgP requires that all ports in the channel belong to the same VLAN or are configured as trunk ports. (Because dynamic VLANs can force the change of a port into a different VLAN, they are not included in EtherChannel participation.)


When a bundle already exists and the configuration of one port is modified (such as changing VLAN or trunking mode), all ports in the bundle are modified to match that configuration.


PAgP does not group ports that operate at different speeds or port duplex. If speed and duplex are changed when a bundle exists, PAgP changes the port speed and duplex for all ports in the bundle.


Full link - http://www.cisco.com/en/US/products/hw/switches/ps663/products_tech_note09186a0080094713.shtml


I checked the Best Practices for 6500 IOS as well and it says the same thing.


Thanks for the correction Sarah, rated your post. Hope you keep posting to this forum to keep us all on our toes !


Jon

sarahr202 Sat, 11/29/2008 - 16:53
User Badges:
  • Bronze, 100 points or more

Thanks a lot for your reply Jon and being courageous enough to say" we are wrong".

I really appreciate the way you all answer my stupid, weird questions.

Few weeks ago, I posted question about establishing etherchannel consisting of links in different vlans.

I understand that such a scenario would cause communication problems .

Mr Edison from cisco system responded to my post. He even performed the lab and the result was etherchannel still forms though ports are in different vlans

thanks a lot!

sarahr202 Sat, 11/29/2008 - 20:21
User Badges:
  • Bronze, 100 points or more

I was wondering about pagp that dynamically modifies the parameters such as vlans,duplex,speed.

I was just wondering about changing the vlans of the links in etherchannel,

We usually put interface/port in vlan statically.

Let say we have etherchannel of two-link

between two switches,sw1 and sw2.

Port f0/1 and port f0/2 on sw1 and sw2 are in vlan 1. Now, we change the vlan and put f0/1 in vlan 2. Does it mean pagp will put another link/port (f0/2) in vlan 2 as well?

I wish i could perform a lab to find it out.

thanks alot and have a good night!

Giuseppe Larosa Sun, 11/30/2008 - 04:36
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Sarah,

thanks for your "stupid" questions as you see you are providing your own valuable contribute to the forum.

And forgive us if we sometimes answer using our own experience and understanding that can be wrong as in this case.


Best Regards

Giuseppe

Correct Answer
Jon Marshall Sun, 11/30/2008 - 08:06
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Sarah


Okay things are not as cut and dried as they seem. You inspired me to do some testing so i dug out 2 3550 switches from the garage and after a quick excursion to get crossover cables (you never have them when you need them) i ran the following test -


I connected the 2 switches with an etherchannel on int fa0/10 & int fa0/11. The ports were set to 10Mbps full. I then


1) Changed int fa0/10 on sw1 to run at 100. The etherchannel on both sides marked the port as down.


2) Change int fa0/10 on sw2 to run at 100. The etherchannel on both sides marked the port as suspended.


At no time did changing the anything on fa0/10 effect the settings on fa0/11.


Not sure of any conclusion we can come to yet


1) Maybe there is an order i need to make the changes in

2) The document i posted previously applied to 4500/6500, perhaps it's different with them but unfortunately i don't have either in my garage :-)



Test output

===========


1) All interfaces on both SW1 & SW2 are running 10Mbps full

===========================================================


SW1#sh etherchannel sum

Flags: D - down P - in port-channel

I - stand-alone s - suspended

H - Hot-standby (LACP only)

R - Layer3 S - Layer2

u - unsuitable for bundling

U - in use f - failed to allocate aggregator

d - default port


Number of channel-groups in use: 1

Number of aggregators: 1


Group Port-channel Protocol Ports

------+-------------+-----------+-----------------------------------------------

1 Po1(SU) PAgP Fa0/10(P) Fa0/11(P)


SW1#


2) Changed int fa0/1 on SW1 from 10Mbps full to 100Mbps full

============================================================


SW1#sh etherchannel summ

Flags: D - down P - in port-channel

I - stand-alone s - suspended

H - Hot-standby (LACP only)

R - Layer3 S - Layer2

u - unsuitable for bundling

U - in use f - failed to allocate aggregator

d - default port


Number of channel-groups in use: 1

Number of aggregators: 1


Group Port-channel Protocol Ports

------+-------------+-----------+-----------------------------------

1 Po1(SU) PAgP Fa0/10(D) Fa0/11(P)


SW1#



3) Changed int fa0/10 on SW2 from 10Mbps to 100Mbps full

========================================================


SW1#sh etherchannel summ

Flags: D - down P - in port-channel

I - stand-alone s - suspended

H - Hot-standby (LACP only)

R - Layer3 S - Layer2

u - unsuitable for bundling

U - in use f - failed to allocate aggregator

d - default port


Number of channel-groups in use: 1

Number of aggregators: 1


Group Port-channel Protocol Ports

------+-------------+-----------+-----------------------------------------------

1 Po1(SU) PAgP Fa0/10(s) Fa0/11(P)


==========================


I was intending to run the test you proposed in your last response and i will when i get the chance.


Jon

sarahr202 Sun, 11/30/2008 - 19:38
User Badges:
  • Bronze, 100 points or more

thanks for your reply!

Look like cisco press book was wrong again.

I believe it is very unlikely for pagp to change the vlan dynamically for all the ports in etherchannel.

Thanks alot Jon! I just envy your having two 3550 switches in the garage!

Jon Marshall Mon, 12/01/2008 - 02:10
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Sarah


Its not definite one way or the other. That Cisco press book does seem to be a bit misleading at times but the Cisco documentation for 4500/6500 suggests PaGP will dynamically modify ports as well.


So unless/until someone can test this on a 4500/6500 setup probably best to just say we are not sure either way.


I'll keep the switches handy as i'm sure you'll be posting more questions :-)


Jon

sarahr202 Mon, 12/01/2008 - 16:16
User Badges:
  • Bronze, 100 points or more

You bet! I will be posting questions :-)

cnadl Wed, 12/03/2008 - 05:10
User Badges:

hello jon,


i'm interested with the test. are these port configured as auto/desirable or on? i'm not so sure if this will affect the test result or not for i don't have an environment. there's another possibility like you said, difference per platform.


and we know that if changes was made on po interface, the physical interface will change their configuration as well. maybe this is what cisco press wanna tell us?


a.d.l

Jon Marshall Wed, 12/03/2008 - 05:29
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Hi


The etherchannel mode was configured as desirable as i think setting it to on would not have been a good test.


"and we know that if changes was made on po interface, the physical interface will change their configuration as well. maybe this is what cisco press wanna tell us?"


Yes, i thought about this too. The documentation does seem to be suggtesting though that a change to any one of the individual ports will propogate to all the other ports.


I suspect it may be platform specific but wouldn't want to say for sure without testing.


Jon

Actions

This Discussion