To ensure that VCS supports the Oracle Restart feature, the user needs to tune the VCS parameters so that it does not attempt a failover action when the Oracle grid infrastructure attempts to restart the resources on the same node. For VCS to take correct decision, the ToleranceLimit resource attribute needs to be tuned accordingly.

Oracle Restart feature uses the CHECK_INTERVAL and RESTART_ATTEMPTS attributes to determine the monitoring interval and restart attempts if the resources are unavailable.

For example, when ASM instance becomes unavailable on a node, the grid infrastructure tries to restart the ASM resource, however VCS detects that the application is unavailable and reports the resource as faulted and tries to failover the service group to another cluster node. To avoid such scenarios, we need to ensure that the VCS policies do not take effect until Oracle completes performing its functions.

You need to modify the VCS resource ToleranceLimit attribute such that ToleranceLimit * MonitorInterval is greater than (CHECK_INTERVAL *RESTART_ATTEMPTS) + MonitorInterval .

If Oracle's attribute values for ASM resource is CHECK_INTERVAL = 1 (in seconds) and RESTART_ATTEMPTS = 5 (no of restart attempts), and the VCS configuration values for the ASMInst resource has MonitorInterval set to 60, then set the ToleranceLimit to 2 based on the below calculations,

2 (ToleranceLimit) * 60 (MonitorInterval) = 120
1 (CHECK_INTERVAL) * 5 (RESTART_ATTEMPTS) + 60 (MonitorInterval) = 65

The following commands can be used to set the Attribute ToleranceLimit :