Oracle® Virtual Networking QDR and EDR Host Drivers for Oracle VM 3.4.1 Release Notes

Exit Print View

Updated: September 2016
 
 

Known Issues

The following known issues are for this release of Oracle Virtual Networking host drivers for Oracle VM 3.4.1.

Bug ID
Description
16338092
For a PVI vNIC connected to a VM running any OS other than RHEL 6u2, a PVI vNIC will stay in up state even if the vNIC is actually set to down on the Oracle Fabric Interconnect. This problem occurs if you first create the vNIC in a PVI, then set the vNIC interface to none, set the vNIC interface back to the PVI, then attempt to set the vNIC to down.
After this series of events, the PVI vNIC will not be set to down state unless the vNIC is in a RHEL 6u2 VM on an Oracle VM 3.4.1 server.
16338065
With iSCSI storage, a problem prevents the deletion of iSCSI LUNs from the array. If you attempt to delete an iSCSI LUN from the iSCSI array, after refreshing Oracle VM Manager the LUN will still be displayed and the port connected to the LUN will be put into an error state. This problem also occurs with physical NICs, so it appears that this is a problem in Oracle VM Manager. Be aware to explicitly delete an iSCSI LUN from Oracle VM Manager. This requirement exists for vNICs and physical NICs alike.
16338007
A problem with Oracle VM hosts prevents passing traffic (even pings) across a PVI vNIC if the MTU is set to 64000 on a RHEL 5u8 virtual machine. During testing, an MTU of 9138 K passed traffic in all cases. Using a 9 K MTU on a PVI vNIC should be used.
16337950
For Oracle VM hosts, a problem in the host drivers prevents a newly added LUN from being automatically recognized.
You can workaround this problem by issuing the xsigo-scan -r -a command from the Oracle VM server.
16335100
If an event causes failover or restart of the OpenSM InfiniBand subnet manager, multicast traffic is delayed for approximately seven seconds while the failover or restart occurs.
16334399
Disks can be prevented from being displayed on a Linux host (or VM running in an Oracle VM server). This problem manifests in either of the following ways:
  • If LUN Masks are not configured on the Oracle Fabric Interconnect and the vHBA is present, if you then create the LUN and map it to storage, the disk is not automatically displayed.

    In this situation, you can work around the problem by running a target rescan on the Oracle Fabric Interconnect and then running the xsigo-scan -r -a command on the Oracle VM host to make the LUN visible.

  • If LUN Masks are configured on the Oracle Fabric Interconnect, but the vHBA is not already present. then if you create the vHBA and attach it to storage, the disk is not automatically displayed.

    In this situation, you can work around the problem by running a target rescan on the Oracle Fabric Interconnect and then running the xsigo-scan -r -a command on the Oracle VM host to make the LUN visible. Or, in cases where the vHBA is not already present and is added later, you will need to run the xsigo-scan -r -a command on the Oracle VM 3.4.1 host to make the LUN visible.