ixgben pNICs in a teaming may experience network issues.

Products

• VMware vSphere ESXi

Issue/Introduction

The vmxnet3 backend may experience "Hang" in TX side as the following log specifies:

YYYY-MM-DDTHH:MM cpu2:743150)Vmxnet3: 26636: <VM_NAME>,<MAC_Address_of_Nic>, portID(100663565): Hang detected,numHangQ: 1, enableGen: 473

Cause

ixgben driver will not clean/re-allocate resources (packets inflight) when it receives link down/up event passively, because VF connection would be impacted by such behavior.
While vmxnet3 is pending on driver to send out all those packets and complete them, vmxnet3 will be monitoring them but hit timeout (5 ~ 10s) finally.

Resolution

There is a known issue affecting all inbox/async ixgben drivers on ESXi. This may cause network instability related to physical NICs (pNICs), NIC teaming, and VMXNET3 adapters.

The issue typically self-corrects within 5–15 seconds.

Workaround if the issue does not self-resolve:

 

• For Predictable Port Down Events (e.g., planned maintenance):
  You can proactively take down the pNIC using the following ESXi command:

   

  esxcli network nic down -n vmnicX

• For Unpredictable Port Down Events (e.g., physical switch port or cable issues):
  Unfortunately, there is no effective workaround. The issue must be mitigated by addressing the underlying hardware or connectivity instability.