| United States-English |
|
|
|
![]() |
HP Integrity Virtual Machines: Release Notes > Chapter 3 RestrictionsVirtual Networking |
|
This section lists the release notes associated with virtual networks. The hpvmnet command displays the status of the vswitches, including the mode. The vswitches are always in SHARED mode. This indicates that the vswitch is set to promiscuous mode. Dedicated (non-promiscuous) mode is not used. Host to guest connectivity may not be reliable when using the HP A5506B PCI 10/100Base-TX 4 Port interface for guest networking. When you add a virtual NIC to your guest, Integrity VM checks to make sure the MAC address is unique. By default, Integrity VM makes three attempts (each with a one-second timeout) to determine the validity of the MAC address for the virtual NIC. This process can result in up to ten seconds of delay for each defined virtual NIC. To speed up this processing, add the following tunable to the /etc/rc.config.d/hpvmconf configuration file:
Where n is the number of attempts (1 to 30). The default is 3. A value of 1 or 2 increases performance at the risk of missing a response from a slow NIC. Do not set the HPVMMACADDRFRAMES tunable to 0. Setting this tunable to 0 can result in a virtual NIC having a duplicate MAC address with another virtual or physical NIC.The first guest to boot obtains the MAC address, so a guest with a duplicate MAC address would fail to boot. To boost virtual network performance, create additional vswitches and allocate them across guests. Integrity VM does not support running APA in the guest. You can run APA on the VM Host. APA can be configured on the VM Host to provide a highly available LAN for the vswitch (APA in active/passive mode) or to increase the bandwidth of the vswitch LAN (APA active/active mode). Before you stop APA, halt the vswitches associated with it. Use the hpvmnet -h command to halt the vswitches. If you do not halt the vswitch first, the hpvmnet command reports an incorrect MAC address for the vswitch. Vswitches must not be connected to network devices that are set to promiscuous mode. Do not run applications like tcpdump on the VM Host on interfaces that are used for virtual switches. TCP Segmentation Offload (TSO) is turned off by default in HP-UX. HP recommends that you leave it turned off on both the VM Host system and on HP-UX guests. This applies to both the virtual network interface cards in the guest and any physical network interface cards in the VM Host that are used by vswitches. Turning on TSO results in guest network interruption. To verify whether TSO is turned on, enter the following command:
Where n is theVM Host interface, as displayed by the hpvmnet. For example, to verify that lan0 is on, enter the following command:
To turn TSO off on lan0, use the following command:
Do not use the hpvmnet command to create a virtual switch that is associated with a VLAN port on the VM Host (that is, a LAN created with lanadmin -V). This “nested VLAN” configuration is not supported. When you use a VLAN for communication between guests and the VM Host, turn off transmit checksum offload (CKO) on the network interface used by the vswitch. On the VM Host, identify the PPA of the network interface for the vswitch using the hpvmnet command. For example:
Check the status of the transmit CKO using the following command:
In this example, the VLANs are configured over the vswitch vmlan4. This vswitch is created on PPA 4 on the VM Host. To turn off CKO on PPA 4, enter the following command on the VM Host:
It is necessary to restart the vswitch when:
For information about how to restart vswitches, see the HP Integrity Virtual Machines Installation, Configuration, and Administration manual. When you restart a vswitch, it is not necessary to restart the guests using the vswitch. |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||