OCR is like a Windows registry that stores all configuration information related to the cluster and RAC database. It is also a public configuration, that is, multiple nodes share the same configuration information. Therefore, this configuration should be stored on the shared disk. This article mainly describes the OCR of the cluster and the amnesia caused by OCR based on Oracle 10g RAC.
I. OCR features
Similar to the Windows registry, it is used to store all configuration information related to the cluster and RAC database.
Shared by multiple nodes. Therefore, it can only be stored on shared disks. Supports single disk and image storage. Usually 100 MB-1 GB.
In Oracle 10 Gb, it can only be stored on bare devices or ocfs file systems, as well as nfs and gfs file systems. Oracle 11g can be directly stored in asm.
The configuration of the entire cluster and RAC database must be maintained in OCR. In other words, it is like importing, exporting, modifying, and updating key values in the windows registry.
In general, the configuration information in OCR is automatically updated with the use of tools. Such as SRVCTL, DBCA, OEM, and NETCA.
OCR configuration and maintenance usually include OCR verification, backup, view OCR content, add and remove OCR files, relocate, and repair OCR files.
An OCR disk can contain only two Primary OCR disks and one Mirror OCR disk. The two OCR disks are mirrored to prevent single point of failure (spof) on the OCR disk.
Note: The cluster generally refers to clusterware, And the RAC database is a database based on the cluster.
Ii. Content contained in OCR
OCR usually contains the following content:
Node member information
Database instances, nodes, and other Mappings
ASM
Resource configuration information (vip, services, etc)
Service characteristics)
Process Information in the Oracle Cluster
Third-party application information controlled by CRS
Iii. OCR Process
Because OCR is stored in shared storage, each node in the Cluster accesses a copy protected by OCR cache in its memory through the local OCR process. In addition, the OCR
All operations must ensure OCR content integrity. Therefore, not all nodes can operate OCR disks during ORACLE Clusterware operation. Only one OCR Process
Read and Write operations by using the optical character recognition (OCR) in the storage. This node is called the OCR Master node. This process is responsible for refresh its own local cache and other nodes in the Cluster.
. In other words, the OCR client queries a local OCR copy through the local OCR process. When the client needs to update the OCR, the local OCR
The process interacts with the OCR Master, which plays the role of reading and writing OCR files.
# Below is a two-node RAC. It can be seen that the ocr master node was first in the first node, then transferred to the second node, and then to the first node.
# This is because the ocr master node is transferred when Node 1 is closed or the cluster on node 1 is unavailable.
Oracle @ bo2dbp:/u01/oracle/crs/log/bo2dbp/cssd> cat ocssd. log | grep "master node"
CLSS-3001: local node number 1, master node number 1
CLSS-3001: local node number 1, master node number 1
CLSS-3001: local node number 1, master node number 1
CLSS-3001: local node number 1, master node number 1
CLSS-3001: local node number 1, master node number 2
CLSS-3001: local node number 1, master node number 1
Oracle @ bo2dbp: ~> Grep-I "master node" $ ORA_CRS_HOME/log/bo2dbp/cssd/ocssd. log | tail-1
CLSS-3001: local node number 1, master node number 1
OCR client applications include: Oracle universal Installer (OUI), SRVCTL, Enterprise Manager (EM), DBCA, DBUA, NetCA, and VIPCA ).
In addition, OCR maintains and manages the dependency and status information of various application resources defined in CRS, especially Database, Instance, Services, and node applications.
The OCR configuration file is named ocr. loc, which is located in/etc/oracle/ocr. loc in Linux.
Attached OCR Structure