11-17-2020 11:58 PM
Hello Experts,
I am currently facing an issue with etherchannel loosing connectivity with VMWare ESXi 6.7 host server. This is the new setup.
So I have got 2 C9300 series switches (C9300-48UXM-A and C9300-48P-A) stack together. And 2 gig ports from C9300-48P-A are part of etherchannel connecting to VMWare ESXi host.
Problem :- I loose the connectivity to the server every now and then. I need to shut and no shut one of the member ports in the channel to restore the connectivity but connectivity last only for couple of minutes.
I neither see any related error in logs nor seeing port going in err-disable mode. port-channel and both the ports status shows UP at the time I loose the connectivity. Below is the configuration on the switch stack. I have also attached output of "show version" command and interfaces status from the time I lost the connectivity to the server.
Please advise as to what I am missing here.
interface Port-channel1
switchport trunk native vlan 999
switchport trunk allowed vlan 20,50,199
switchport mode trunk
spanning-tree portfast trunk
end
interface GigabitEthernet2/0/38
switchport trunk native vlan 999
switchport trunk allowed vlan 20,50,199
switchport mode trunk
channel-group 1 mode on
spanning-tree portfast trunk
end
interface GigabitEthernet2/0/45
switchport trunk native vlan 999
switchport trunk allowed vlan 20,50,199
switchport mode trunk
channel-group 1 mode on
spanning-tree portfast trunk
end
11-18-2020 12:26 AM
Post the complete output to the following commands:
sh controller e Gi2/0/38 sh controller e Gi2/0/45 sh platform pm interface-flaps Gi2/0/38 | b LinkNeg sh platform pm interface-flaps Gi2/0/45 | b LinkNeg
11-18-2020 04:36 AM
11-18-2020 11:00 AM
I've reviewed the output.
There is nothing wrong with the links -- not a Layer 1 issue.
I'm going to say that the issue is at the VM side.
11-18-2020 08:42 PM
Hello
I will re-visit ESXi side configuration for any correction. Thank you for reviewing the output.
11-18-2020 03:17 AM
- Check (networking related) logs on ESX too , to possibly get further insights.
M.
11-18-2020 04:49 AM
Check the port-channel config and hashing algorithm on VMware
11-18-2020 06:33 AM
The interfaces have:
492 unknown protocol drops
So, can you make sure that the ESXi site is configured the same as the switch (mode on)?
If that does not help, try using both interfaces without any Portchannel configuration. Sometimes VMs and Portchannels don't play well together.
HTH
11-18-2020 07:13 AM
Hello
I do not see any specific option to enable/disable "mode on" in ESXi host but port-channel/interfaces don't come up if I set the mode to either active (LACP) or auto (PAgP) on stack switch. I have attached the screenshot of ESXi Nic teaming configuration which was configured following the VMWare official document.
Regarding unknown protocol drops, I have read one of the discussion in Cisco community that these drops occurs when the interfaces in port-channel, are configured with trunk and that such errors can be ignored. Correct me if that understand is wrong.
Can this be an issue of the code version on the stack switch? Or port-channel connectivity with ESXi host on the stack switches doesn't work well?
BTW, I configured both interfaces without port-channel and the connectivity got restored. But is this a best practice to have dual port connectivity with ESXi host without port-channel? And will I get failover/load balancing capabilities?
Just for information, I have same NIC teaming on ESXi and port-channel/interfaces configuration on C2960 stack switch for other location and it is working just fine.
11-18-2020 11:06 AM
Hi,
BTW, I configured both interfaces without port-channel and the connectivity got restored. But is this a best practice to have dual port connectivity with ESXi host without port-channel? And will I get failover/load balancing capabilities?
Yes, as long as the NIC teaming/failover is set up correctly on the ESXi hosts, the failover from one interface to other will work just fine. We have Dell servers with ESXi hosts configured this way and when we test by shutting down one of the interfaces on the server or the switch, the failover works just fine.
HTH
11-18-2020 11:50 AM
Hello
Disable DTP (switchport nonegotiate) on the PC trunk that should deal with the unknown protocol drops.
Does the server support LACP is so try using it instead of a static PC.
11-18-2020 08:39 PM
Hello
DTP is already disable. I am not seeing any option in ESXi to enable LACP on server but when I enable it on switch side, port-channel doesn't come UP and in the logs, I can see the error stating "LACP is not supported on remote port". So, I guess, LACP is not supported on server.
11-18-2020 04:26 PM - edited 11-18-2020 04:37 PM
11-18-2020 08:41 PM
Hello
I have followed thew same article to create the vSwitch and build the NIC teaming on server.
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