Veritas Volume Manager 5.0.1 Release Notes, HP-UX 11i v3, First Edition, November 2009

Additionally, the vxdmpadm setattr arraytype array_type command sets the
attribute for all array types derived from the given array_type.
Support for ALUA JBOD Devices
The Device Discovery Layer (DDL) now provides improved support for Just a bunch
of disks' (JBOD) devices to include Asymmetric Logical Unit Access (ALUA) JBOD
devices. DMP now provides immediate basic support for any ALUA-compliant array,
however full support still requires an ASL. For more information on the supported
arrays, see the Hardware Compatibility List (HCL) at:
http://docs.hp.com/en/5900-0354/5900-0354.htm
NOTE: The DMP feature and its enhancements are available only with the VxVM-full
license.
PFTO disabled in the HP-UX Native Multipathing Devices
By default, the use of Page Fault Time Out (PFTO) is now disabled in the HP-UX native
multipathing devices. As a result, the native multipathing disk I/O can consume additional
service time to complete an I/O successfully. In the case of DMP devices, the use of PFTO is
enabled by default.
Support for LVM version 2 Volume Groups
The LVM version 2 volume groups are now partially supported. VxVM now identifies and
protects LVM version 2 volume groups. However, the LVM version 2 volume groups cannot
be initialized or converted.
Distributed Volume Recovery
In a cluster configured using Cluster Volume Manager (CVM), when a node crashes, the
mirror recovery is initiated by the CVM master. Prior to this release, the CVM master
performed all recovery I/O tasks. Starting from this release, the CVM master can distribute
recovery tasks to other nodes in the cluster. Distributing the recovery tasks is desirable in
some situations so that the CVM master can avoid an I/O or CPU bottleneck.
When the distribution of volume recovery is enabled, the CVM master uses the round-robin
method to distribute recovery tasks to other nodes in the cluster. By default, all of the nodes
in the cluster can participate in volume recovery. You can also exclude particular nodes
from the volume recovery.
To enable the distribution of volume recovery, add the keyword distribute in the /etc/
default/vxrecover file. You can also specify the distribute keyword while running
the /etc/default/vxrecover file. To disable the distribution of volume recovery, remove
the distribute keyword from the /etc/default/vxrecover file. If the feature is not
enabled, the CVM master performs the resynchronization tasks locally as it was performing
in the previous release. The distributed volume resynchronization functionality does not
depend on the disk group version, and so works with older version of disk groups.
New Features in This Release 9