Veritas Storage Foundation 5.1 SP1 Cluster File System Administrator"s Guide (5900-1738, April 2011)
■ Forcefully unregister other nodes (preempt) as members of this active
SFCFS cluster
In short, the CP server functions as another arbitration mechanism that
integrates within the existing I/O fencing module.
Note: With the CP server, the fencing arbitration logic still remains on the
SFCFS cluster.
Multiple SFCFS clusters running different operating systems can
simultaneously access the CP server. TCP/IP based communication is used
between the CP server and the SFCFS clusters.
About preferred fencing
The I/O fencing driver uses coordination points to prevent split-brain in a VCS
cluster. By default, the fencing driver favors the subcluster with maximum number
of nodes during the race for coordination points. With the preferred fencing
feature, you can specify how the fencing driver must determine the surviving
subcluster.
You can configure the preferred fencing policy using the cluster-level attribute
PreferredFencingPolicy as follows:
■ Enable system-based preferred fencing policy to give preference to high
capacity systems.
■ Enable group-based preferred fencing policy to give preference to service
groups for high priority applications.
■ Disable preferred fencing policy to use the default node count-based race
policy.
See “How preferred fencing works” on page 35.
See “Enabling or disabling the preferred fencing policy” on page 117.
How preferred fencing works
The I/O fencing driver uses coordination points to prevent split-brain in a VCS
cluster. At the time of a network partition, the fencing driver in each subcluster
races for the coordination points. The subcluster that grabs the majority of
coordination points survives whereas the fencing driver causes a system panic
on nodes from all other subclusters. By default, the fencing driver favors the
subcluster with maximum number of nodes during the race for coordination
points.
35Storage Foundation Cluster File System architecture
About I/O fencing