|
|
|
|
|
CRS application resources are similar to VCS resources. Each component controlled by CRS is defined by an application resource, including databases, instances, services, and node applications.
Unlike VCS, CRS uses separate resources for components that run in parallel on multiple nodes.
Resources are defined by profiles, which are similar to the attributes that define VCS resources. The OCR contains application resource profiles, dependencies, and status information.
CRS node applications are the resources CRS uses to manage RAC databases in a high availability environment.The installation of CRS creates a set of node applications by default. CRS uses these application resources to manage Oracle components, such as databases, listeners, and virtual IP addresses.
During installation, a nodeapps object is created as a container for all these application resources. This enables starting and stopping application resources as a group action.
The voting disk is a heartbeat mechanism used by CRS to maintain cluster node membership. Voting disk data exists on a shared raw volume or a cluster file system that is accessible to each node. This requires approximately 20 MB of disk space.
The ocssd processes of CRS provides cluster node membership and group membership information to RAC instances. On each node, ocssd processes write a heartbeat to the voting disk every second. If a node is unable to access the voting disk, CRS determines the cluster is in a split brain condition and panics the node.
Because the voting disk is a critical component and single point of failure in a CRS cluster, the storage for the voting disk uses either the Oracle redundancy feature or third-party mirroring and multipathing features to prevent interruptions to the cluster. Placing the storage objects for the voting disk under VCS control ensures that the voting disk is in place before the ocssd process is started (through resource dependencies) and provides notification in the event the voting disk fails.