Network connectivity for VM is down for 30 seconds during failover of uplink.

Products

• VMware NSX

Issue/Introduction

• Link aggregation control protocol is set for the VMs and vmkernels on the distributed switch , causing the network failover from one uplink to another taking around 30-60 seconds.
• ESXi and Virtual machines losing network connectivity during the time of failover.
• Shutting down one uplink is not immediately bringing down the interface on the switch end.
• LACP is set to passive on the vCenter end and active on the physical network side.
• From packet captures, we are noticing packets leaving the host immediately once the uplink goes but no response from the physical switch side 
• On the physical side, will see ethernet in suspended state.
• Will notice below entry on the switch side:

 

Resolution

Workaround:

1. Set the LACP timeout mode to fast , PDUs are sent every second
2. Verify the reason for uplink not going down immediately on the physical switch side.