A problem causes the state of checksumming on a PVI vNIC to be the opposite of the state you specifically set. For example, if you create a PVI vNIC with checksum enabled, then check the PVI vNIC on the ESXi 5.5.2 host, the checksum flag shows disabled. And if you disable checksumming on the PVI, the checksum flag shows enabled on the PVI.

In a previous version, after enabling IP over IB (IPoIB) on the ESXi host, IPoIB adapters are not listed when you list network adapters with esxcfg-nics -l. This problem is present in the current release of Oracle Virtual Networking host drivers for ESX hosts.

An internal function call (GET_COALESCE) in the host driver caused ESXi 5.5 servers to crash to pink screen of death (PSOD) when attaching an Oracle Virtual Networking vNIC adapter as an uplink to a distributed Vswitch through the ESX web client.

In earlier versions of ESX host drivers, vNICs or vHBAs could take too long to come up during a server boot up, which prevented ESX autodeploy functionality in some build versions of ESXi hypervisor.

The Oracle Virtual Networking host drivers contain an IP over InfiniBand (IPoIB) module in the host drivers archive that you download and install. A problem prevented this IPoIB module from loading automatically on ESXi hosts. The module now loads automatically, but you can manually enable or disable the IPoIB module by following te instructions in Enabling or Disabling IP Over IB.

Deleting a vNIC was successful, but the vNIC was still displayed in VMware vSphere in a down state.

With XgOS 3.7.2 and ESXi 5.x hosts running Oracle Virtual Networking driver version at least version 5.2.1, a problem caused server profiles to transition to up/down state. When the problem occurred, vNICs and vHBAs were put into up/indeterminate state, and host connections to storage and networking resources were lost.

With Compellent storage arrays, after a failover completes and VM load is manually rebalanced, a problem in the vHBA driver software can cause a flood of ABORT and RSCN messages. When the messages occur, the vHBA is also attempting a rescan (RSCN). This series of events can cause a VM kernel panic.

Oracle 1-Gbps vNICs connected into a vSwitch on an ESXi server running at least version 5.1 were erroneously reported as 10 Gbps. This issue was cosmetic only, and the actual traffic speed on the vNIC was as configured.