cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
9494
Views
8
Helpful
11
Replies

Discussion about etherchannel with LACP enabled on cisco switch

jiangcaixia
Level 1
Level 1

Hi guys,

I want to have some discussions about the etherchannel with lacp enabled.

From LACP protocol, an aggregation group binds the ports with the same speed, duplex and linkstate. Therefore, for ports with speed 100 and speed 1000 will belong to the different aggregation groups for traffic.

Cisco switch can be configured LACP for etherchannel group with "channel-group x mode active/passive". I am wondering whether etherchannel group can contain multiple aggregation groups(such as different groups with different link speeds). Or, should all the ports belonging to the same etherchannel group be compatible? With same speed, duplex ...? Is this requirement needed when we configure channel-group?

I just did an experiment: if two ports are configured for the same channel group with different link speeds, it will throw the error
"5d08h: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet0/11, changed state to down
5d08h: %EC-5-CANNOT_BUNDLE2: Gi0/11 is not compatible with Gi0/12 and will be suspended (speed of Gi0/11 is 100M, Gi0/12 is 1000M)". Is this error related to the above requirement?

And one more question, if this requirement is true, when the link speed of one port is changed, what is the status of this channel-group? with other ports binded and this port suspended? How about the traffic?

Can anyone help answer these questions? Thanks in advance!

Regards,
Caixia

11 Replies 11

Kevin Dorrell
Level 10
Level 10

Hi Caixia, you seem to have the feature well.  An EtherChannel group cannot contain multiple aggregation groups, and all members must have the same characteristics.  (In fact "aggregation group" can be considered the same thing as "etherchannel group")

As you observed, if you try and gat a 100 Mbps link to join a group tghat has 1 Gbps members, it will be suspended.  This will also happen if you change the link speed (or many of the other properties) of the link.  As fas as the traffic is concerned, you might not even notice because all the traffic will failover to the remaining links pretty well instantly.

This should not, however, discourage you from configuring LACP.  I have found from bitter experience that the "on/on" configuration is actually quite dangerous, and should be avoided if at all possible.  (With on/on, if you make a cabling error, you can send you network into meltdown.)

Hope this helps.

Kevin Dorrell

Luxembourg

Hi Kevin,

Thanks a lot for your comments!

Sorry that I still have one confusion. I think an important feature of LACP is to automatically bind the ports with different capabilities to different aggregation groups. It means that it will automatically adjust the group members when the link capability changes.

But for cisco switch, all these aggregation groups (channel groups) are determined when we configure them with "channel-group x mode active/passive" commands. And when the link state of one port in the channel group is not compatible with others, it will suspend and all the traffic failover to the remaining links in the group. So, this failed link will just exit this aggregation group but not participate a new group. And it can reparticipate the group when its link status resumes. I think the above LACP feature is not well used.

I learn from your comments that the important feature of LACP in cisco switch is that it can load balance when the link fails in the channel group. It is better than "on" mode. When "on" mode is configured, the traffic will be messed after a link fails.

Is there something that I understand wrong? Look forwards to your reply! Thanks:)

Regards,

Caixia

Hi Caixia,

Link Aggregation Control Protocol (LACP) is part  of an IEEE specification (802.3ad) that allows you to bundle several  physical ports together to form a single logical channel. LACP allows a  switch to negotiate an automatic bundle by sending LACP packets to the  peer. It performs a similar function as Port Aggregation Protocol (PAgP)  with Cisco EtherChannel.

Modes of LACP :

LACP trunking supports four modes of operation, as follows:

  • On: The link aggregation is forced to be formed without any  LACP negotiation .In other words, the switch will neither send the LACP  packet nor process any incoming LACP packet. This is similar to the on  state for PAgP.

  • Off: The link aggregation will not be formed. We do not send  or understand the LACP packet. This is similar to the off state for  PAgP.

  • Passive: The switch does not initiate the channel, but does  understand incoming LACP packets. The peer (in active state) initiates  negotiation (by sending out an LACP packet) which we receive and reply  to, eventually forming the aggregation channel with the peer. This is  similar to the auto mode in PAgP.

  • Active: We are willing to form an aggregate link, and initiate  the negotiation. The link aggregate will be formed if the other end is  running in LACP active or passive mode. This is similar to the desirable mode of PAgP.

The above modes only detremines how the negotiation takes place in terms of bundling ports in an ether channel.

Any of the modes stated above wouldnt be affecting tha traffic passing through links of the ether channel.

Traffic distribution in a ether channle depends on load balancing over ether channel and has nothing to do with the modes .

I hope the above answeres your query.

You can refer to teh follwoing document for more clarification:

http://www.cisco.com/en/US/tech/tk389/tk213/technologies_configuration_example09186a0080094470.shtml

** please rate if you find the content useful **

Hi Swati,

Thanks for your input!

I know that the modes have no relations with the traffic. The traffic is determined by load balancing policies. I just want to comfirm the behaviors of LACP for cisco switch, the comments written in update#3. Thanks!

Best,

Caixia

I use cisco switch 3560 and from its configuration guide for etherchannel it says:

"When you configure one end of an EtherChannel in either PAgP or LACP mode, the system negotiates with the other end of the channel to determine which ports should become active. Incompatible ports are put into an independent state and continue to carry data traffic as would any other single link. The port configuration does not change, but the port does not participate in the EtherChannel."

But when I changed the link speed of one port, it will show the port is suspended. And it seems all the traffic sent to that port are blocked. So I am confused:

1. What's the suspended status? Is the traffic blocked in this status?

2. What's the way that cisco switch deals with link speed change? Put these incompatible ports into blocked status or use them as the single links?

Can someone give some comments? Thanks in advance!

Regards,

Caixia

Caixia,

Incompatible ports are put into an independent state and continue to carry data traffic as would any other single link. The port configuration does not change, but the port does not participate in the EtherChannel."

But when I changed the link speed of one port, it will show the port is suspended. And it seems all the traffic sent to that port are blocked. So I am confused

I believe that the emphasized statement refers to the situation when the interconnected ports on neighboring (i.e. different) switches are considered by LACP as incompatible. It does not talk about what happens if you make two member ports of the same port-channel on the same switch incompatible. Such a situation is actually not an issue of LACP at all - the LACP negotiates EtherChannels between different switches, not between different ports of a same switch.

Does this make sense? Feel welcome to ask further.

Best regards,

Peter

Hi Peter,

Thanks for your reply!

From your comments, I learn that etherchannel with lacp enabled can be configured on neigboring switches.Does it mean that the ports on different switches can be binded into a logic channel group? (such as ports on switch A and ports on switch B binding to a channel group connect to switch C)

Can you give me some instructions or links about how to configure such a case? I don't find the descriptions on the configuration guide. Thanks!

And I guess from your reply that the situation (the port will be suspended for the channel group after its speed changes) is normal? Is this right? Can you comfirm this situation? And please give me the explanation about what is the suspended status? Thanks very much!

Best,

Caixia

Etherchannels work on a 1 to 1 basis. If you want switches A and B to talk to switch C, each will need to have a seperate port channel built on C. stacked switches count as a single logical switch.

This assumes A and B have links to C

! Switch A

int po 1

desc To switch C

int range g0/1 - 4

channel-group 1 mode active

! Switch B

int po 1

desc To switch C

int range g0/1 - 4

channel-group 1 mode active

! Switch C

int po 1

desc To switch A

int po 2

desc To switch B

int range g0/1 - 4

channel-group 1 mode active

int range g0/5 - 8

channel-group 2 mode active

Thanks, pkaretnikov.

Etherchannel balances the traffic load across the links in a channel. So, in the above case, switch C can only load balance individually in channel group 1 or channel group 2. I wonder whether there is a technique that binds the ports in switch A and ports in switch B both to one aggregation group in switch C? Therefore, when all ports in A fail, C can load the traffic to B. I heard vPC(virtual portchannel) can do it. Is it right? Can cisco switch support it? Thanks!

Best,

Caixia

I believe that what you are looking to do is only on higher end switches such as the Nexus line. It also can create less than optimal switching situations. If you don't have a very good reason for doing that I would recommend against implementing it.

Removed redundant post

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco