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

All vPars and VM guests running on a V6.3 VSP must have the V6.3 VirtualBase bundle installed
to take advantage of the PCI OLR capability on the VSP.
The guest has more than 32 NPIV vHBAs that are backed by the I/O card that are
considered for replacement. PCI OLR is not supported on guests having more than 32
devices backed by the I/O card that are considered for replacement.
OR
Guest Instance (Guest ID) vswitch LanX: The Guest has more than 32 vNICs that are backed
by the IO card being considered for replacement. PCI OLR is not supported on guests
having more than 32 devices backed by I/O card that are considered for replacement.
A PCI OLR operation or a CRA on the VSP is not supported if any of the active vPars or VM guests
have more than 32 impacted IO devices. The operation can only be performed after bringing
down the vPar or VM guest.
You must ensure that no guest has more than 32 NPIV vHBAs or more than 32 AVIO LAN interfaces
backed by the same I/O card on the VSP to take advantage of the PCI OLR capability on the VSP.
The guest user space daemon is not running; see the HPVM documentation for possible
workarounds
With vPars and Integrity VM V6.3, a new HPVM guest user space daemon (/opt/hpvm/bin/
hpvmgud) is delivered as part of the VirtualBase bundle. The daemon gets launched as part of
the guest boot, and is a pre-requisite for performing a CRA on the guest.
A PCI OLR operation or a CRA on the VSP is not supported if the HPVM guest user space daemon
is not running or is unresponsive in any of the active vPars or VM guests with impacted IO.
The following checks and actions can be performed to ensure that the HPVM guest user space
daemon is operational:
• Check all the process on the guest OS and see that hpvmgud is running (ps -aef | egrep
-i "hpvmgud"
• View syslog for guest user daemon log messages.
• Remove the hpvm entry from /etc/inittab and run /sbin/init q and then start the
guest user daemon manually by running /opt/hpvm/bin/hpvmgud without any options.
The guest CRA timed out, see HPVM documentation for possible workarounds.
For CRA timed out error and possible workaround, see Section 11.4.9 (page 192).
The guest CRA failed due to an HPVM internal error, check HPVM documentation for
possible workarounds.
This error is issued when the guest is in a state where PCI OLR components on the VSP are unable
to communicate with the necessary components within the guest. This includes all guest states when
boot or shutdown is in progress. In these states, several guest daemons and kernel operations will
not be able to operate reliably.
The solution is to retry the PCI OLR operation on VSP when all guests are in a stable state - either
not running at all or has completed start up operations.
Could not suspend the driver in one of the guests.
Error: prep_replace:/usr/sbin/olrad.d/hpvmdio driver script Failed !
A PCI OLR Suspend operation failed on the VSP in at least one of the guests. Try the PCI OLR
Suspend operation once again on the same device on the VSP.
168 PCI OLR support on VSPs