4.0.0 HP PolyServe Software for Microsoft SQL Server Administration Guide (T5392-96056, March 2010)

The default policies (NOFAILBACK for the virtual host and AUTORECOVER for the monitor) cause the
ClusterPulse process to fail over the virtual host to a backup node. The virtual host remains on the
backup node until a “healthier” node becomes available, at which point the virtual host will fail over
to that node.
You can use the Timeout and Failure Severity attribute to change the failover/failback behavior. There
are three settings: NOFAILOVER, AUTORECOVER, and NOAUTORECOVER.
NOFAILOVER. When the monitor probe fails, the virtual host does not fail over to a backup node.
This option is useful when the monitored resource is not critical, but is important enough that you want
to keep a record of its health. To use this option, you must also specify IGNORE as the Event Severity
on the Scripts tab.
AUTORECOVER. If the virtual host policy is AUTOFAILBACK, when the probe fails, the ClusterPulse
process automatically begins failover of the associated virtual host to a backup node. Failback occurs
in accordance with the policy setting for the virtual host (either AUTOFAILBACK or NOFAILBACK).
NOAUTORECOVER. When the probe fails, the ClusterPulse process automatically begins failover of
the associated virtual host to a backup node. The monitor is then disabled on the original node. If the
virtual host policy is AUTOFAILBACK, you will need to manually reenable the monitor before failback
can occur. This option is useful when integrating HP PolyServe Software with a custom application,
where certain application-specific actions must be taken before the failback can occur.
Service Priority: The service priority is used when the ClusterPulse process fails over virtual hosts.
Service priorities are natural numbers, with 0 (zero) representing the highest priority and higher
numbers representing lower priorities. If multiple failures prevent ClusterPulse from placing a virtual
host on a node where all of its associated services are available, ClusterPulse next looks for a server
where the associated service with the highest priority is available. By default, all service monitors
have a priority of 0.
Probe Type: The probe type is set to single-probe. A single-probe monitor performs the probe function
only on the node where the monitor instance is active. The monitor instance is activated on the node
where the associated virtual host is activated, and the probe takes place on that node. The monitor
instances on other nodes are marked as “standby” on the Management Console.
If the virtual host fails over to a backup node, the monitor instance on the original node becomes
inactive and the probe is no longer run on that node. Matrix Server activates the virtual host on the
new node, which causes the monitor instance on that node to change status from “standby” to “active.”
The monitor then begins probing on that node and the Start script starts the application. If the monitor
instance cannot be activated on the new node, the virtual host will fail over to another backup node
if possible.
Scripts
The Scripts tab lets you customize a monitor with Post Start, Post Stop, and Recovery scripts. You can
also configure the event severity and script ordering for the monitor.
Configure Virtual SQL Servers64