09-22-2017 12:05 AM - edited 03-08-2019 12:07 PM
Hi,
I am having difficulty implementing what should be an easy task.
I have a single ESXi 6.5U1 host with 2 NICs teamed in a single vSwitch. The load balancing is set to “route based on IP hash”. As part of my troubleshooting I have also disabled beacon probing (Net.MaxBeaconsAtOnce=0).
The NIC team connects to a single Cisco 2960 switch. Both interfaces are up and the host is reachable.
To avoid Mac flapping on the switch (due to the NIC teaming), I am trying to configure an Etherchannel on the switch. My config below (both switch ports are configured identically):
interface GigabitEthernet0/6
switchport trunk native vlan 1000
switchport mode trunk
channel-group 2 mode on
end
interface GigabitEthernet0/7
switchport trunk native vlan 1000
switchport mode trunk
channel-group 2 mode on
end
interface Port-channel2
switchport trunk native vlan 1000
switchport mode trunk
end
EtherChannel Load-Balancing Configuration:
src-dst-ip
Group Port-channel Protocol Ports
------+-------------+-----------+-----------------------------------------------
2 Po2(SU) - Gi0/6(P) Gi0/7(P)
The post channel comes up on the switch but the ESXi host is not reachable. As soon as I shut one of the ports, the comms is restored. I can’t figure out what I am doing wrong – any ideas?
Thanks
Solved! Go to Solution.
09-22-2017 02:31 AM
Got it sorted.
Facepalm.
The load balance algorithm set on the vSwitch does not propagate to the management network.
https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1022751
Thanks for confirming my etherchannel config on the switch is correct and guiding me to look on the ESXi side!
09-22-2017 12:27 AM
09-22-2017 12:53 AM
Thanks, I have attached my server side config - I can't see anything wrong :(
Note that for these sreenshots, I have removed the Etherchannel on the switch to indicate that the correct interfaces form part of the NIC team (remember that with the Etherchannel configured, I can only reach the ESXi host if I shut one interface on the Cisco switch).
Can you spot anythign issue here? Thanks again
09-22-2017 12:29 AM
Hi
sounds like something is off on the server side , your switch config is correct cehck this guide may help you
09-22-2017 12:55 AM
Thanks - I have added my server config above (can't spot the problem unfortunately).
I have worked through the article you posted already and I have configured my environment as close as possible to this guide ... still no luck however.
09-22-2017 12:58 AM - edited 09-22-2017 12:58 AM
have you tried LACP instead as an option see if its having an issue being forced on
09-22-2017 01:01 AM
Yes, with LACP (mode access) the the Etherchannel does not come up at all. I believe this might be because I am using a vSwitch - not a distributed switch (not an option in my environment).
09-22-2017 01:02 AM
09-22-2017 01:07 AM
09-22-2017 01:16 AM
09-22-2017 02:04 AM
Thanks Mark - I tried that (no native vlan defined) but with the same result - no comms to the host.
09-22-2017 02:31 AM
Got it sorted.
Facepalm.
The load balance algorithm set on the vSwitch does not propagate to the management network.
https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1022751
Thanks for confirming my etherchannel config on the switch is correct and guiding me to look on the ESXi side!
09-22-2017 02:33 AM
09-22-2017 02:38 AM
09-22-2017 03:24 AM
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