09-16-2013 06:30 AM - edited 03-07-2019 03:29 PM
Hello,
I have two 4510R-E switches both running Sup6-E for the supervisor. We've noticed some odd issues with traffic flowing between switches and I wanted to start off by getting a sanity check on the configuration. Below is the config (identicle on the other switch except for the interface numbers) as it stands now. I think during my next maintenance window, I will change the mode from desirable to on, but I'm not sure if that will make any difference. So, I guess, my question is, what is the best configuration for the etherchannel between these two switches?
interface Port-channel2
description *** Uplink between 4510R-E switches ***
switchport
switchport mode trunk
switchport nonegotiate
interface GigabitEthernet2/15
description *** Uplink to BERN01PORS2 4510R ***
switchport mode trunk
switchport nonegotiate
channel-group 2 mode desirable
interface GigabitEthernet2/16
description *** Uplink to BERN01PORS2 4510R ***
switchport mode trunk
switchport nonegotiate
channel-group 2 mode desirable
Solved! Go to Solution.
09-16-2013 10:09 AM
Hi Mark,
I think during my next maintenance window, I will change the mode from desirable to on
I wouldn't recommend to do so; using PAgP or LACP for negotiate the EtherChannel is the safer approach.
As far as I see, there's nothing wrong with your EtherChannel config.
What kind of media is used for interconnecting the switches? If it's fiber, do you have UDLD enabled on those links?
There are some useful show-commands to check channel status, distribution among the links and status changes (timestamps):
show etherchannel [
show etherchannel [
show pagp [
Perhaps you could check if everything looks stable, of course you can post the output if you have any doubts.
Hope that helps
Rolf
09-16-2013 10:09 AM
Hi Mark,
I think during my next maintenance window, I will change the mode from desirable to on
I wouldn't recommend to do so; using PAgP or LACP for negotiate the EtherChannel is the safer approach.
As far as I see, there's nothing wrong with your EtherChannel config.
What kind of media is used for interconnecting the switches? If it's fiber, do you have UDLD enabled on those links?
There are some useful show-commands to check channel status, distribution among the links and status changes (timestamps):
show etherchannel [
show etherchannel [
show pagp [
Perhaps you could check if everything looks stable, of course you can post the output if you have any doubts.
Hope that helps
Rolf
09-16-2013 04:03 PM
channel-group 2 mode desirable
If the EtherChannel is ONLY between two Cisco boxes, then I'd use "mode on".
09-16-2013 04:59 PM
I agree with Rolf. be careful when using "Mode On" as I have seen this take down a network when one of the links comes unbundled for whatever reason. You may also look into using UDLD to ensure your links function properly.
09-16-2013 05:05 PM
The config is fine and what I always use and have never had any issue with PAGP channel negotiation . Changing it to on will make no difference. Myself I use desirable non-silent as the channel option . LACP is the other option . I also manually prune off any unneeded vlans across the trunk links .
09-16-2013 08:17 PM
Thanks everyone for the input. I was pretty sure the config was sound, but I needed a sanity check. I ran the show commands, and everything looked ok as well. I think we've identified the issue to be something between our UCS system/Fabric Interconnects and the 4510 switches as the issue only appears on systems running within the UCS environment.
Thanks again, and I will post in the UCS community should I have any quesitons regarding that.
Cheers,
Mark
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide