Cisco LACP Bond, ports are Downing themselves?

[ASCitnetwork]

Hey Guys,

Hoping you can help with this, we have multiple 4 port LACP Port-channels that are direct connected to Dell PowerEgde 610s.  Over time, random ports in the port-channels will Down themselves (show as (D)) and wont come back up unless we manually shut/no shut them.

Any idea what could be causing this?

Here is a sample config of one of the port-channels:

switchport trunk encapsulation dot1q
switchport trunk allowed vlan 30,35
switchport mode trunk
spanning-tree portfast

And one of the interfaces in the port-channel:

interface GigabitEthernet1/0/25
switchport trunk encapsulation dot1q
switchport trunk allowed vlan 30,35
switchport mode trunk
channel-group 25 mode active
spanning-tree portfast

[Reza Sharifi]

Hi,

"spanning-tree portfast" should not be applied to trunk ports. That is for access ports.

Also, do you have the same issue if you use mode "on" (no LACP)?

HTH

[ASCitnetwork]

Hey Riza,

Thanks for the quick reply, could the Spanning-tree be causing any issues?

We have not tried changing it to channel-group mode on, I will see if we can test that,

Thanks

[Reza Sharifi]

Hi,

Not sure if it is STP but if mode "on" does not help, for testing you can just trunk the ports (no Portchannel) from one of the servers and see if that makes any difference. If the servers are VM host, simply trunking them to the switch should work as sometimes Dell servers don't play nicely with Cisco switches running Portchannles.

HTH

[Deepak Kumar]

Hi, 

I saw you applied a command "spanning-tree portfast". Remove the command and test again. 

Second: Can you confirm which mode is used in Vswitch to configure Link-Aggregation?

Regards,

Deepak Kumar

[ASCitnetwork]

Hey Deepak,

I have removed the spanning-tree command, unfortunately they interfaces randomly go down over days/weeks so there isn't a solid way to test.

On the vswitch, we have it on Proxmox.  It is configured as an OVS bond and is set for LACP (balance-tcp) mode.

[Deepak Kumar]

Hi, 

Can you try to remove the port-channel 25 and try again with below configuration?

Do with default reset to connected interface. 

no interface port-channel 25

default interface GigabitEthernet1/0/25

default interface GigabitEthernet1/0/XX

!

interface port-channel 25

switchport trunk encapsulation dot1q
switchport trunk allowed vlan 30,35
switchport mode trunk
!

interface GigabitEthernet1/0/25
switchport mode trunk
channel-group 25 mode active
no shut

!

interface GigabitEthernet1/0/XX
switchport mode trunk
channel-group 25 mode active
no shut

Regards,

Deepak Kumar

[ASCitnetwork]

Hey Deepak,

We can try rebuilding the port-channel once we get scheduled downtime again.  These port-channels host our Virtual Machines so they are high important

[gs.skills]

Hello,
before reconfiguring or shut/no shut, you should verify interface status for err-disable status:
show interface status err-disabled

Regards, Guillaume

[Reza Sharifi]

You know with ESXi hosts, you can simply use trunks using a couple of physical ports. 

This way, you would know if the Portchannel is the issue or not.

HTH 

[ASCitnetwork]

Hey Reza,

Do you know if the same holds true for ProxMox boxes?

We are going to bring up a test environment,

Thanks

[Reza Sharifi]

Hi,

As long as you are using VMware, it should work the same way.

HTH

[Reza Sharifi]

It is really a good idea to have a test environment, so you test multiple scenarios and not affect the production VMs. Once you have a solid solution, you can then deploy it in production.

HTH