csumErr) on random Rx queues of particular physical NICs cause this issue
# esxcli network nic stats get -n vmnicX
NIC statistics for vmnic1:
Packets received: 4802380244Total receive errors: 2244
Queue Statistics:
rxq0: totalPkts=1381473796 totalBytes=471928939608 nonEopDescs=0 allocRxBufFail=0 csumErr=1836rxq1: totalPkts=810298599 totalBytes=1168073677028 nonEopDescs=0 allocRxBufFail=0 csumErr=408
Hardware Offloading: This can occur if the NIC’s hardware offload engine improperly validates incoming packets or if the packets themselves were corrupted during transmission across the physical switch fabric.
Non-Harmful Protocols: Certain management or discovery protocols (e.g., Cisco DTP, WLCCP) may use non-standard frame structures that trigger checksum or length errors in the ESXi driver accounting, despite being harmless to the environment.
csumErr continues to increment rapidly, engage the physical networking vendor to perform a packet capture (using pktcap-uw) on the switch port to identify if a specific device or protocol is sending malformed frames.