Enabling Solaris SMF Services to Run Under the Control of Sun Cluster
The Service Management Facility (SMF) enables you to automatically start and restart SMF services, during a node boot or service failure. This feature is similar to the Sun Cluster Resource Group Manager (RGM), which facilitates high availability and scalability for cluster applications. SMF services and RGM features are complementary to each other.
Sun Cluster includes three new SMF proxy resource types that can be used to enable SMF services to run with Sun Cluster in a failover, multi-master, or scalable configuration. The SMF proxy resource types enables you to encapsulate a set of interrelated SMF services into a single resource, SMF proxy resource to be managed by Sun Cluster. In this feature, SMF manages the availability of SMF services on a single node. Sun Cluster provides cluster-wide high availability and scalability of the SMF services.
For information on how you can encapsulate these services, see Enabling Solaris SMF Services to Run With Sun Cluster
You might require Sun Cluster to make highly available an application other than NFS or DNS that is integrated with the Solaris Service Management Facility (SMF). To ensure that Sun Cluster can restart or fail over the application correctly after a failure, you must disable SMF service instances for the application as follows:
-
For any application other than NFS or DNS, disable the SMF service instance on all potential primary nodes for the Sun Cluster resource that represents the application.
-
If multiple instances of the application share any component that you require Sun Cluster to monitor, disable all service instances of the application. Examples of such components are daemons, file systems, and devices.
Note –
If you do not disable the SMF service instances of the application, both the Solaris SMF and Sun Cluster might attempt to control the startup and shutdown of the application. As a result, the behavior of the application might become unpredictable.
For more information, see the following documentation:
- © 2010, Oracle Corporation and/or its affiliates