HP-UX vPars and Integrity VM V6.3 Administrator Guide

Redefining pNICs for HP-UX guests
Changing the hardware address of a vswitch has the same effect as moving a network adapter
from one hardware slot to another on an HP Integrity system. Similar to other HP-UX systems, the
guest file /etc/rc.config.d/netconf must be modified so that INTERFACE_NAME[0] reflects
the new LAN PPA assigned by the HP-UX network driver on the first guest reboot after modification.
At the first reboot, the LAN interfaces configuration fails, as follows:
Configure LAN interfaces ............................. FAIL*
When the guest is running, you can use the nwmgr command to identify the new LAN PPA and
netconf command to modify the new LAN PPA. For example:
# nwmgr
Name/ Interface Station Sub- Interface Related
ClassInstance State Address system Type Interface
============= ======== ============ ======== ========= =========
lan3 UP 0x02636c6E3030 iexgbe 10GBASE-KR
In the preceding example, before the modification, the LAN PPA was 0. The new LAN PPA on the
first boot after the modification is 3. To resolve this, you must bring the guest network down, then
you must change the INTERFACE_NAME[0] from lan0 to lan3. You can then use /sbin/
rc2.d/S340net to restart the guest network. For example:
# /sbin/rc2.d/S340net stop
# ch_rc -a -p "INTERFACE_NAME[0] = "lan3"
# /sbin/rc2.d/S340net start
The guest network begins to function.
Problems with VLANs
When VLANs are configured on the vswitch, the partitioned LAN must have its own set of network
servers to service requests on the VLAN.
If guests start slowly or hang during starting, determine whether the guest network interface is on
a VLAN, and whether the appropriate network services (such as DNS) are set up and available
on the VLAN. You might need to disable some of these network services on the guest before booting
up the guest on a VLAN.
When VLANs are configured on the vswitch and the guests are required to communicate over a
VLAN with a remote node outside the VSP, you might need to set up the physical network
appropriately for the VLAN. For information about configuring VLANs on the switches, see the
product documentation for the physical network adapters.
If TCP/UDP applications have trouble communicating between a guest and the local VSP over a
VLAN, it is possible that the host interface for the vswitch is checksum-offload capable. To resolve
the problem, identify the interface used by the vswitch and run the following command on the VSP
to disable the CKO feature, where 4 is the VSP interface as shown in the hpvmnet command
output.
# nwmgr -s -A tx_cko=off -c lan4
lan4 current values:
Transmit Checksum Offload=Off
Checksum offloading (CKO) is not supported. On most of the physical interfaces that are
not of 10 Gigabyte type, CKO is turned off by default. Consult your interface card documentation
for details.
Turning on CKO can cause host-to-guest connections as well as guest-to-host communication over
a VLAN to fail. If you are receiving failures with host-to-guest connections or guest-to-host
communication using a VLAN, ensure that the CKO is turned off in the host interface driver. If that
does not fix the problem, reboot the vswitch.
To turn off the CKO on the VSP, identify the PPA of the network interface for the vswitch using the
hpvmnet command. For example:
# hpvmnet
270 Troubleshooting