While upgrading Cisco infrastructure FI's. One Fabric Interconnect was rebooted and it looks as if the secondary vmnic didn't pickup the traffic.

Products

• VMware vCenter Server

Issue/Introduction

When rebooting a Fabric Interconnect, the vNICs of integrated UCS servers can report in an online state (link up) prior to the uplink ports of the Fabric Interconnect being configured.

Cause

This can lead to traffic failing over to the vNICs reporting as online and traffic loss being encountered as there is no available uplinks for the traffic to traverse upstream from the recently rebooted Fabric Interconnect.

Resolution

This is a bug reported by Cisco, please engage Cisco TAC for the full resolution.

https://bst.cisco.com/quickview/bug/CSCwf86353

A workaround can be performed to set a delay in ESXi to not fail traffic back to this link for a particular amount of time, the default value is 100ms.
Cisco will need to provide a timeframe that this FI will be expected to not pass traffic, a command can be used to delay ESXi failing back to the connection that is being reported as "link up".
The following command can be run to set the delay with the time entered being displayed in ms.

esxcli system settings advanced set -o /Net/TeamPolicyUpDelay --int-value <time in milliseconds that Cisco provides>