A physical link or vNIC/PVI-vNIC that is assigned to DV switch as uplink, will not be available after ESXi-host reboot (VMware bug).

Workaround: Reassign physical link or vNIC/PVI-vNIC as uplinks to the respective DV switch ports.

The net-ib-cm VIB is not present with ESXi 5.1 with version 5.3.1 of Oracle Virtual Networking host drivers for ESXi hosts. As a result, esxcli-update operations going from ESXi 5.0 releases to ESXi 5.1 or later will fail because of the absence of the VIB in the older versions of host drivers. Upgrading from ESXi 5.1 to another release equal to, or greater than, ESXi 5.1 should not experience this problem.

Workaround: Use a remastered ISO when updating to the latest version of host drivers.

Assume you have successfully created a PVI vNIC on a PVI Cloud. If you then remove that PVI vNIC, rename it, and attempt to add it back to the PVI cloud, the vNIC is put into up/fail state.

Workaround: Do either of the following:

If you add vNICs and vHBAs with the same name to an ESXi host, the first virtual I/O device comes up. When you add the second virtual I/O device, it comes up in a failed state.

Workaround: Ensure that vNICs and vHBAs have unique names.

Consider the situation where EMC storage is directly connected to the Oracle Fabric Interconnect and is available to an ESXi server running at least version 5.1, and also running PowerPath. If FibreChannel frames get dropped, the ESXi server might crash and display a pink screen of death (PSOD). This issue originates in PowerPath code.

Workaround: No workaround exists for this issue.

Consider the situation where an ESXi server is running at least version 5.1, is connected to storage through multipath vHBAs, and is running PowerPath 5.7. If you delete the active vHBA, the server might crash and display a pink screen of death (PSOD).

Workaround: Reboot the server, then stop traffic on the active vHBA that you want to delete by running the set vhba name down command before deleting the active vHBA.