Veritas Volume Manager 5.1 SP1 Administrator"s Guide (5900-1506, April 2011)

■ Any failures that require a configuration change must be sent to the master
node so that they can be resolved correctly.
■ As the master node resolves failures, all the slave nodes are correctly updated.
This ensures that all nodes have the same view of the configuration.
The practical implication of this design is that I/O failure on any node results in
the configuration of all nodes being changed. This is known as the global detach
policy. However, in some cases, it is not desirable to have all nodes react in this
way to I/O failure. To address this, an alternate way of responding to I/O failures,
known as the local detach policy, was introduced.
The local detach policy is intended for use with shared mirrored volumes in a
cluster. This policy prevents I/O failure on any of the nodes in the cluster from
causing a plex to be detached. This would require the plex to be resynchronized
when it is subsequently reattached.
The local detach policy is supported for disk groups that have a version number
of 120 or greater.
For small mirrored volumes, non-mirrored volumes, volumes that use hardware
mirrors, and volumes in private disk groups, there is no benefit in configuring
the local detach policy. In most cases, it is recommended that you use the default
global detach policy.
The choice between local and global detach polices is one of node availability
versus plex availability when an individual node loses access to disks. Select the
local detach policy for a diskgroup if you are using mirrored volumes within it,
and would prefer a single node to lose write access to a volume rather than a plex
of the volume being detached clusterwide. i.e. you consider the availability of your
data (retaining mirrors) more important than any one node in the cluster. This
will typically only apply in larger clusters, and where a parallel application is
being used that can seamlessly provide the same service from the other nodes.
For example, this option is not appropriate for fast failover configurations. Select
the global detach policy in all other cases.
In the event of the master node losing access to all the disks containing log/config
copies, the disk group failure policy is triggered. At this point no plexes can be
detached, as this requires access to the log/config copies, no configuration changes
to the disk group can be made, and any action requiring the kernel to write to the
klog (first open, last close, mark dirty etc) will fail. If this happened in releases
prior to 4.1, the master node always disabled the disk group. Release 4.1 introduces
the disk group failure policy, which allows you to change this behavior for critical
disk groups. This policy is only supported for disk groups that have a version
number of 120 or greater.
Administering cluster functionality (CVM)
Overview of clustering
432