Tx Hang occurs whenever a vNIC is hot-added to a DarkTrace OVF deployed VM

Products

• VMware vSphere ESXi
• VMware vSphere ESX 8.x

Issue/Introduction

A VCF Administrator observes a network outage on the ESXi host when attempting to hot-add a VNIC to a DarkTrace VM:

2026-06-16T20:20:54.197Z In(182) vmkernel: cpu40:2097211)Vmxnet3: 18934: Tx completion timeout exceeded for tq 0
2026-06-16T20:20:54.197Z In(182) vmkernel: cpu40:2097211)Vmxnet3: 19295: Device42-rc-19.05.10.eth0,00:50:56:##:##:##, portID(100663374): Hang detected,numHangQ: 1 enableGen: 106
2026-06-16T20:20:54.197Z In(182) vmkernel: cpu40:2097211)Vmxnet3: 19309: portID:100663374, qid: 0, next2TX: 97, next2Comp: 79,next2Write:244, ringSize: 512 inFlight: 2, delay(ms): 8122,  txStopped: 0
2026-06-16T20:20:54.197Z In(182) vmkernel: cpu40:2097211)Vmxnet3: 19326: portID: 100663374, sop: 79 eop: 80 enableGen: 106 qid: 0, pkt: 0x45b97f006dc0
2026-06-16T20:20:54.197Z In(182) vmkernel: cpu7:2097442)Performing Live coredump: vmxnet3-initiated

Environment

ESXi 8.0.3
DarkTrace VM deployed from OVF

Cause

This is a bug within vmxnet3 that is triggered when a vNIC is hot-added to a VM when that VM's GuestOS memory has not been mapped. In a mirrored packet situation, if the fast suspend/resume that occurs as a result of the hot-add vNIC function takes too long, this could result in a Tx hang on the physical NIC, which can impact anything else using that NIC including storage traffic.

Resolution

This issue has been identified as a bug in the vmxnet3 code. A fix will be released in the future once it has a release vehicle. This KB will be updated when that code fix is available.