9800 WLC connected to Meraki MX losing connectivity

[ToddTaylor]

I have an HA pair of 9800-L-C wireless Controllers running 17.12.4.  Our corporate SSID is sent out port te0/1/0 to our core switches and is working fine.  Our Guest SSID sends traffic out of port te0/1/1 to a Meraki MX105.  The Meraki runs DHCP and is connected to a separate ISP dedicated for guest.

Once every other week or so, the connection will drop.  I have to do a shut/no shut on the te0/1/1 port to restore it.  I have a script automated through SolarWinds that runs when it detects 0 bps traffic on that port for 30 seconds. 

I can find no rhyme or reason as to why the traffic stops.  It happens at random times and dates.  Sometimes a few days apart, and sometimes weeks.  1:00am on a Saturday or 2:00pm on a Tuesday. 

Has anyone seen anything like this before?  TAC hasn't been very helpful.

interface TenGigabitEthernet0/1/0 (Corporate traffic working fine)
switchport mode trunk
negotiation auto
no snmp trap link-status
service-policy output AutoQos-4.0-wlan-Port-Output-Policy
!
interface TenGigabitEthernet0/1/1 (Guest port with the issue)
switchport trunk native vlan 14
switchport trunk allowed vlan 14
switchport mode trunk
speed 1000
no negotiation auto
no snmp trap link-status
service-policy output AutoQos-4.0-wlan-Port-Output-Policy

[Scott Fella]

Not really knowing the full configuration, have you tried to use the config analyzer to validate your configuration?  Also, can the issue be on the MX side, do you see any logs on that port?

[ToddTaylor]

I've had Meraki look at their side as well.  They stop seeing DHCP requests come through and the WLC says it's getting no response when it sends.

I'll check out the config analyzer.

[Scott Fella]

There are no logs on either end like if the port negotiation changed or anything, have you tried to leave it at auto? It would be interesting if you connect interface TenGigabitEthernet0/1/1 to the core as an access port and then push that vlan to the MX.  You can use a bogus vlan like 666 so that the traffic is isolated and see if you run into that issue again. Just thinking out loud what I would try.