Broadcom BCM57414 vmnics missing in ESXi after host reboot

Products

• VMware vSphere ESXi

Issue/Introduction

The Broadcom BCM57414 NetXtreme-E network interface card (NIC) may fail to appear in the ESXi network inventory (esxcli network nic list) despite appearing on the PCI bus.

Symptoms:

• After rebooting an ESXi host, Broadcom BCM57414 NetXtreme-E 10Gb/25Gb RDMA Ethernet Controller ports (vmnics) may disappear from the hypervisor's network inventory.

• While the device remains visible on the PCI bus (via lspci), it is not listed in esxcli network nic list, and the bnxtnet driver fails to initialize the device.

• Log entries in vmkernel.log or boot logs indicating firmware reset failures:

  WARNING: bnxtnet: dev_reset_chip:3848: [0000:xx:00.x] FW reset failed!!
WARNING: bnxtnet: bnxtnet_attach_device:498: [0000:xx:00.x] dev probe ph2 failed(status: Failure)
WARNING: bnxtnet: hwrm_snd_fw_msg:492: ... HWRM cmd too long, cmd_len= 160, max_len supported= 128
WARNING: bnxtnet: bnxtnet_init_queue_info:983: [0000:d8:00.0 : 0x452113040000] HWRM_FUNC_CFG failed from firmware

Environment

VMware ESXi

Cause

The issue is caused by a firmware-level hang or crash on the NIC. During the host boot sequence, the bnxtnet driver attempts to initialize and reset the chip. If the firmware is in a non-responsive state, the reset command fails (FW reset failed!!), causing the driver probe to abort. Consequently, the vmnic is never created in the ESXi stack. This can occur due to transient hardware lockups or internal state inconsistencies that persist through a standard warm reboot. A firmware hang or "zombie" state occurs when the installed driver attempts to send commands that exceed the firmware's supported buffer length (160 bytes requested vs. 128 bytes supported). In some cases, the adapter is a generic/OEM variant that does not accept standard Broadcom or server-vendor firmware updates, preventing the necessary firmware reconciliation.

Resolution

To resolve this issue, a full hardware reset of the network adapter is required. Follow these steps:

1. Cold Power Cycle:
   

   • Place the host in Maintenance Mode.
     

   • Shut down the ESXi host.

   • Physically disconnect all power cables from the server for at least 60 seconds, or perform a "Cold Reset" / "Full Power Cycle" via the server's IPMI/BMC interface.

   • Reconnect power and boot the host.

2. Verify vmnic Status:
   

   • Once the host is back online, run the following command to check if the NICs are detected: esxcli network nic list

3. Verify Alignment: Ensure the driver and firmware versions are aligned to the 234.x branch or higher.

4. Reseat Hardware (If Step 1 fails):
   

   • If the vmnics are still missing, power down the host and physically reseat the BCM57414 AIOM module/PCIe card in its slot.

5. Hardware Vendor Engagement:
   

   • If a cold power cycle and reseat do not restore the NICs, contact your hardware vendor to investigate potential hardware failure or to check for further out-of-band firmware updates.

Additional Information

• Ensure driver and firmware versions are aligned according to the certified recipes: IO Devices Compatibility Matrix

• For additional support with downloads, see Download Broadcom Products and Software.

• KB: Network cards not detected on an ESXi host