ASM Single point of failureIt is often the user's best practice to ask how to avoid an ASM single point of failure. As with other file systems or volume managers, the failure of an ASM instance can cause the RDBMS instance running above to be unavailable. However, OS file system or volume manager failure typically accompanies operating system crashes, and ASM Restart does not require a server restart. The best way to solve an ASM single point of failure is to run ASM in a RAC environment that is highly available through the cluster. ASM can continue to run on other nodes if a node restarts or an RDBMS restarts. ASM failure is a soft crash recovery faster than a traditional file system or volume manager.
ASM InstallationIf the server or an ASM instance on a node manages only one RDBMS, it is sufficient to suggest that the ASM and RDBMS share a single oracle_home. However, for an ASM to manage the storage of several RDBMS instances, it is recommended that ASM be installed in a separate oracle_home (Asm_home).
| Translator Note: From the development of Oracle, even if ASM only corresponds to a set of RDBMS also recommended oracle_home separate, 11g ASM Independent to the GI, so it is recommended to separate the good, easy maintenance, in our production environment I am independent open storage 10205 ASM Run 10204 Rdbms. This makes it possible to use the kfed repair function. |
Note: In a RAC environment, if you perform an Add node operation and implement a standalone Asm_home,addnode operation, you also need to run on asm_home. Starting with Oracle Database 10gRelease 2, Oracle Universal Installer (OUI-DBCA) and DBCA have been enhanced to enable standalone Oracle_ Create an ASM instance in home OUI-DBCA can be installed with the following features: 1. You can configure the database to use ASM to manage Storage 2. You can create an ASM instance, do not create a database 3.ASM is already running a database on the host, and DBAs can use the EM migration Tool to move the database forward to ASM
initializing an ASM instance of ORACLE 10GStart $ORACLE_HOME/BIN/DBCA to configure ASM instances. Translator Note: The process of configuring ASM instances is simple and not explained in detail here. See the original book P50
Upgrading ASM SoftwareASM instances upgrade from 10.x to 11.x upgrades can be manually upgraded and Dbua upgrade using two methods. Regardless of which method you use to upgrade 11g, Oracle_home is created. Note that upgrading is simply a software upgrade that requires the use of all 11g features in a disk group upgrade, which can then be processed by a disk group upgrade.
Manual Upgrade1. Update ORATAB2 using the ASM oracle_home directory. Copy the Init.ora and make the appropriate modifications, such as diag and dump locations. In a RAC environment, the ASM home configuration in the cluster needs to be modified and can be accomplished using the SRVCTL command: Srvctl modify asm-n racnode1-i +asm1-o \/opt/oracle/app/product/11.1/asm-p I Nit+asm1.ora needs to execute this command on all cluster nodes participating in ASM. 3. After the instance is started, grant Sysasm permissions to Sysuser 4. For Windows systems, you need to use Oradim to rebuild Windows services. Translator Note: If the previous step is ASM upgrade to 11g steps, 10g upgrade is actually very simple, upgrade the software can be.
Dbua UpgradeASM upgrade to 11g you can use Dbua,dbua to perform the following steps: 1. Copy the password file in the new Asm_home and rebuild the init.ora/spfile2. Unix/linux on modify Oratab or on Windows using Oradim Rebuild Service 3. Grant Sysasm permissions for Sysuser to start 11goracle_home from Dbua. Note: If the installation ASM is detached oracle_home (asm_home), you need to start Dbua under asm_home/bin/instead of the database Oracle_home Translator Note: Dbua Upgrade graphics Step reference P56
Patching ASM10g ASM does not support ' rolling update ', ' rolling upgrade patchs feature only supports Oracle cluster and Dataguard segments. In Oracle DATABASE 10g, if you are installing a patch set or "nonrolling upgrade safe" patches, all ASM instances in the cluster must stop, and if you use separate asm_home, ASM should be patched in the same way as Rdbmshome. This also includes CPUs (critical patch updates). The ASM rolling upgrade is supported from 11g, and this section describes how to perform an ASM roll-fix upgrade.
