PowerEdge: Broadcom BCM57414 intermittent disconnect issue in NSX-T environment
Summary: Dell PowerEdge server with a Broadcom BCM57414 PCIe adapter could encounter intermittent network-disconnected problem in a VMware NSX-T environment.
Symptoms
Dell PowerEdge server with BCM57414 DP 10G/25G adapter (DPN: H3T3V) configured as the uplinks of an NSX-T virtual switch might encounter intermittent network-disconnected problems.
This issue causes all virtual machines hosted on the problematic server to lose their ability to communicate with external devices through both overlay and underlay networking. Simultaneously, all external equipment is unable to communicate with the virtual machines, applications, and services running on that server.
The problematic NIC port of BCM57414 is still active, which means no failover has been triggered. As a result, the network traffic does not automatically resume while the issue persists. To recover the traffic, either manually disable the problematic NIC port or reboot the server.
Cause
This is caused by a BCM57414 Firmware issue.
TX traffic is routed to the loopback port due to a bad VEB entry.
Resolution
Apply Broadcom NetXtreme-E firmware 22.92.06.10 or newer to resolve this issue.