Alarm "Edge NIC out of Receive Buffer" on BME(Bare Metal Edge)

Products

• VMware NSX

Issue/Introduction

• There are alarms being generated for Bare Metal Edge for interfaces going out of Receive Buffer.
• Below error is seen in the BME /var/log/syslog file       

2000-12-12T04:45:36.756Z test-edge NSX 711555 - [nsx@test comp="nsx-edge" s2comp="nsx-monitoring" entId="xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx" tid="123456" level="WARNING" eventState="On" eventFeatureName="edge_health" eventSev="warning" eventType="edge_nic_out_of_receive_buffer"] Edge NIC fp-eth3 receive ring buffer has overflowed by 2.678912% on Edge node xxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx. The missed packet count is 16 and processed packet count is 348.

Environment

VMware NSX 

Cause

• If CPU utilization and traffic volume are both low but you still see high rx_misses or truncated packets, it often points to an MTU mismatch—oversized frames on a smaller-MTU NIC will be dropped (yet still consume RX descriptors), so verify end-to-end MTU settings.
• MTU(Maximum Transmission Unit) mismatch between the BME(Bare Metal Edge) and Physical switch
• When the MTU on the BME is set to lower than the Physical Switch, the BME is not able to process the jumbo packets eventually filling up the Receive buffer

Resolution

• Make the MTU consistent across all the devices. 
• Set the MTU to either 1700 or 9000 based on your Network configuration and device's capabilities. 
• Update the relevant Tier-0 gateway interface with correct MTU by navigating Go to Networking > Tier-0 Gateway > Edit the Gateway > Expand Interfaces > Select the Interface > Edit Interface > Modify MTU > Save Changes > Save Tier-0 Gateway
• Alternatively, you can update the Uplink profile used by the BME and set the MTU from System > Fabric > Profiles and edit the relevant uplink profile