I did a 2-port Gig EtherChannel between a Cisco 6509 and an Extreme 8810 last month. On the Cisco side, you need to specify LACP as the link aggregation protocol, not PAgP:
channel-group 1 mode active
channel-group 1 mode active
I'm comfortable with the Cisco side of things.
On the extreme (after downloading the command ref guide), i'm looking at:
enable sharing [group #] grouping [ports] lacp
Does the group number need to correlate with the Cisco channel-group number?
I know with cisco it doesn't matter what the channel group number is . I believe it is just a number and shouldn't matter .
In ExtremeXOS or ExtremeWare command syntax, what you are calling [group #] above is actually a port number, which is used to define the logical port identifier for the link aggregation group (LAG) in the Extreme switch.
This port number must be one of the physical ports in the grouping of ports that makes up that particular LAG; but it does not necessarily have to be the first port. For example,
enable sharing 3:9 grouping 3:9-3:12 lacp
defines logical port 3:9 and is made up of physical ports 3:9 through 3:12.
enable sharing 5:7 grouping 3:9-3:12, 5:7-5:10 lacp
defines logical port 5:7 and is made up of physical ports 3:9 through 3:12, and 5:7 through 5:10.
In Cisco IOS, the channel-group number corresponds to the interface Port-channel number. As Glen Grant said, it's just a number. You can pick whichever one you want, as long as it's within the range supported by your switch. (I can pick from 1-256 on my 6509; but only 1-64 on a 3550-12G.)
In Extreme, the "port-group" number is not exactly an arbitrary number, because it's derived from one of the physical ports. Why they do it that way, I don't know. That's just how they do it.
The Extreme LAG logical port number has no correlation with the Cisco channel-group number. Although you can assign some meaning on the Cisco side if you want: âchannel-group 39 mode activeâ or âinterface Port-channel39â to match up with Extreme's LAG port 3:9.
I know this string is old, but I had a similar issues, connecting a 5508 controller to an extreme switch. I found that I could not use LACP at all and ended up using static on the extreme side. So the command would be (in my case):
enable sharing 1:31 grouping 1:31, 1:32
note: no LACP at the end. This works.
I have also had problems in general with the controllers and usually end up setting them to mode on instead of mode active anyway. Go figure.
Is the channel-protocol lacp command really necessary? Isn't LACP implicit in channel-group 1 mode active?
I have just been playing with the channel negotiation protocols in the lab, and I am struggling to understand what useful purpose the channel-protocol command serves.
Strictly speaking, it's not necessary. As you said, "mode active" implicitly uses LACP.
I include "channel-protocol lacp" in the configuration to make extra sure that only a LACP EtherChannel takes place on the specified interfaces. This command prevents you from subsequently adding "channel-group 1 mode auto" or "channel-group 1 mode desirable" to the interface config, because those modes use PAgP. It also creates an error message in the log similar to the following when you try:
â%EC-5-ERRPROT: Channel protocol mismatch for interface Gi1/1 in group 1: the interface can not be added to the channel groupâ
For me, the usefulness of this command is that it makes it just a little bit harder to mess up the config later on. Most of my EtherChannels are between Cisco switches, and I'm using PAgP for those. A year or two from now, when I'm reviewing interface configs and looking to standardize them, and I have forgotten that the switch at the other end is non-Cisco and therefore does not do PAgP, seeing this command line in the config along with the âchannel-group 1 mode activeâ will remind me there's a reason why the mode is different.
We have been trying to do this with an x650 running 12.2.1 and it generally works fine with a 4500, but from time to time it will fail with an Etherchannel misconfig error. The IOS version on the 4500 is 12.2(46)SG. Any ideas on what might be causing this intermittent failure? It's really driving us batty.
Are the trunk type/native vlan/allowed vlans of the Port-Channel and the logical port untagged VID of the aggregator on the Extreme side the same ?
I have a similar issue to Stan and wonder if you can clarify what you are getting at.
We have an 8 port lacp connection between a Cisco 3130 Stack (blade centre switches with stackwise) and an Extreme 8810.
All looks good for a about 25mins, then the 3130 reports:
%PM-4-ERR_DISABLE: channel-misconfig error detected on Gi1/0/18, putting Gi1/0/18 in err-disable state
On all ports and disables them...
It eventually comes back - we have enabled errdisable detect cause all and the 300s retry - but will then go again... No specific error code/state - just the generic channel-misconfig
I need to check the state of the vlan config (if any) on the Extreme 8810 side - there are def. none on the 3130 side... (BTW I'm a server/os guy who wants to get his blade centre trunked up to the core, not a network/switch guy!! ;-)
Thanks for any help