HP P6000 Cluster Extension Software Administrator Guide (5697-0986, June 2011)

Failback option
HP does not recommend auto failback in configurations with P6000 Cluster Extension because the
resource failovers due to storage failure can cause resources to go into an unstable state
(failover/failback might toggle the resource between the nodes).
SLE HA provides the meta-attribute resource-stickiness to determine how much a resource agent prefers
to stay where it is. To disable auto failback, set resource-stickiness to the lowest value compared to
the other resource location constraints.
Migration-threshold
A resource is automatically restarted if it fails. If a restart cannot be achieved on the current node or
it fails to start a certain number of times on the current node, it tries to fail over to another node. You
can define the number of failures for resources (a migration-threshold) after which they migrate to a
new node. If you have more than two nodes in your cluster, the high availability software chooses
the node a particular resource fails over to.
When a P6000 Cluster Extension resource fails, HP recommends configuring your cluster to fail over
the resource without restarting on the local node. To set this preference, set the migration-threshold
to 1.
Disk monitoring
For the situations in which disk access is lost or read/write protection is in effect due to storage fencing,
application monitoring agents, file system agents, or LVM resource agents detect the IO failure. P6000
Cluster Extension does not monitor the disk access status.
RHCS cluster setup considerations
Quorum
In RHCS, the quorum is based on a simple voting majority of the defined nodes in a cluster. To re-form
successfully, a majority of all possible votes is required.
Each cluster node is assigned a number of votes, and they contribute to the cluster while they are
members. If the cluster has a majority of all possible votes, it has quorum (also called quorate);
otherwise, it does not have quorum.
Fencing
Cluster software adjusts the node membership based on various failure scenarios. The concept of
quorum defines which set of nodes continue to define the cluster. To protect data, nodes that do not
have quorum are removed from the cluster. The non-quorate nodes that are removed must be prevented
from accessing the shared resources. This process is called fencing.
HP iLO fencing is one method that can be used with RHCS to restrict cluster node access to shared
resources.
Observe the following guidelines when using HP iLO network configurations with RHCS clusters:
• HP iLO can be connected to the client access network or to a different network, but the network
must be routable.
• HP iLO should not be on the network that is used for cluster communication.
P6000 Cluster Extension features14