After upgrading to ESXi 8.0.3, VMs are facing intermittent packet loss:
The VMs are connected to a vDS using an LACP/LAG connection.
The NICs that are in-use by the impacted VMs are observed to be going down and coming back up with no manual intervention:
vSphere 8.0
The physical NICs used by the VMs are flapping (going down and coming back up) outside of the ESXi.
Looking at the NIC status in CLI esxcli network nic list confirms the NICs are reporting a "Link Status" of "Down", which ESXi does not manage as this is the NIC's physical status as reported by the NIC driver.
The issue can also be isolated by pinging between one of the problematic VMs and another VM that is on the same host and same portgroup/VLAN/subnet while the issue is happening, as this isolates the virtual network from the physical network. Even if NIC flapping is not occurring, if the same host/same portgroup pings are successful the issue is in the physical network and investigation can continue with the physical network team or vendor. See Troubleshooting VMware vSphere ESXi Virtual Machine TCP/IP, ping connection issues for more information.
Investigate the reason for the NIC flapping in the physical network and resolve it.
See Network adapter (vmnic) is down or fails with a Failed Criteria Code for more information about Link Status, even though in this behavior there may not be a Failed Criteria Code logged with the flapping.