Symantec logo

< Previous  | TOC | Index | Next  >  

Relinking the SF Oracle RAC libraries to Oracle

After installing the Oracle base software or Oracle patches, use the configuration program to relink SF Oracle RAC libraries to Oracle.

Relinking for Oracle 10.1.0.5

After you have finished applying patch set for 10.1.0.5, you need to link oracle with SF Oracle RAC libraries.

 To relink the SF Oracle RAC libraries to Oracle 10.1.0.5

  1. Go to /opt/VRTS/install on one node and run following command.

    # ./installsfrac -configure galaxy nebula

  2. Press Enter till you reach Main Menu. Select Option 4 which is for Relink of oracle.
  3. Press Enter to go to "Choose Oracle Version" Menu.
  4. Select option 2, which is for Oracle10gR1.
  5. You are asked to "Choose Task". Select "Relink oracle."
  6. Enter the necessary values for linking oracle with Storage Foundation libraries. Verify the list of all values when prompted.

    To correct values, go back and modify as needed. Once all correct values are displayed in list, Press 'y' to confirm.

  7. The installer validates oracle home, node list and necessary permissions for user name and group. Then, it asks confirmation for stopping crs as "Do you want to continue with stopping crs?". Press 'y' to stop crs. The installer stops CRS.
  8. Exit the installer once crs is stopped successfully.
  9. You can then reboot the nodes to restart crs which is recommended way of restarting CRS in 10gR1 by oracle. After nodes have been rebooted, you need to
  10. The CRS will start after you have made available Private IPs and the OCR and Voting Disk volumes on all nodes. Make sure that CRS_HOME/bin/crs_stat -t shows all oracle resources in ONLINE state.
  11. Then, using Oracle utilities create Listener and Database appropriately. You may refer to "Creating a starter Database" for database creation.

      Note   Note: You MUST create a database if you want to use RAC WIZARD provided by SF Oracle RAC for configuring CFS Mounts and Disk Groups under VCS control on which CRS and database depend. CRS and databases are also configured under VCS by the wizard. Even if you are using manual method for configuring resources under VCS control, its recommended to create database first so as to have an idea of what all resources and their dependencies as well.

  12. Proceed to Configuring Oracle 10g Service Groups after database creation is done.
Relinking for Oracle 10.2.0.2

 To relink the SF Oracle RAC libraries to Oracle 10.2.0.2

  1. Go to /opt/VRTS/install on one node and run following command.

    # ./installsfrac -configure galaxy nebula

  2. Press Enter till you reach Main Menu. Select Option 4 which is for Relink of oracle.
  3. Press Enter to go to "Choose Oracle Version" Menu.
  4. Select option 2, which is for Oracle10gR1.
  5. You are asked to "Choose Task". Select "Relink oracle."
  6. Press Enter.
  7. Enter the necessary values for linking oracle with Storage Foundation libraries. Verify the list of all values when prompted.

    To correct values, go back and modify as needed. Once all correct values are displayed in list, Press 'y' to confirm.

  8. The installer validates oracle home, node list and necessary permissions for user name and group. Then, it asks confirmation for stopping crs as "Do you want to continue with stopping crs?". Press 'y' to stop crs. Installer stops crs as follows.

    Stopping CRS on node1 using: /etc/init.d/init.crs stop ........... Done

    Stopping CRS on node2 using: /etc/init.d/init.crs stop ........... Done

    Checking files ................................................................... Done

    Oracle relinking done successfully

  9. The installer displays LLT and CRS node numberings.

    Checking for node numbering differences ..................... Done

    NODE NAME LLT NODE ID CRS NODE ID

    galaxy 0 1

    nebula 1 2

    Node numbering of LLT and CRS is different. It will be fixed. This step is mandatory for SF Oracle RAC to function.

  10. Press [Enter] to continue.
  11. If LLT and CRS node numberings are not same, then Oracle may not detect some nodes and may detect some nodes in a wrong manner. This step ensures that the two numberings are same. Press Enter to continue.

    Replacing node membership pattern in /etc/VRTSvcs/conf/config/main.cf on

    all the nodes ... Done

    NodeId attribute has been changed only for CVMCluster resource in

    /etc/VRTSvcs/conf/config/main.cf to conform to Oracle's node numbering

    scheme, but if you have any other resource which uses NodeId as an

    attribute, you must change that manually in

    /etc/VRTSvcs/conf/config/main.cf file before starting vcs.

    Replacing node membership pattern in /etc/llthosts on all the nodes .. Done

    Press [Enter] to continue:

  12. The installer prompts you to stop the whole Storage Foundation for Oracle RAC stack and provides the required commands.

    Use the commands to stop the Storage Foundation for Oracle RAC stack.

    Before finishing Oracle 10g Release 2 installation, the SF Oracle RAC stack needs to be shutdown on all the nodes using the following steps.

    Commands to shutdown the stack :

    1) Stop database, crs.

    2) Unmount all VxFS mounts, stop volumes and deport diskgroups.

    3) Stop DBAC with commands :

    /etc/init.d/vcsmm stop (Do not use vcsmmconfig -U);

    /etc/init.d/lmx stop

    4) Stop ODM and GMS with commands :

    /etc/init.d/vxodm stop; /etc/init.d/vxgms stop

    5) Stop VCS with command :

    /opt/VRTSvcs/bin/hastop -local

    6) /opt/VRTS/bin/fsclustadm cfsdeinit

    7) Stop GLM with command :

    /etc/init.d/vxglm stop

    8) Stop/Kill the following processes if they are running :

    had, hashadow, CmdServer

    9) Stop VXFEN with command :

    /etc/init.d/vxfen stop (Do not use vxfenconfig -U)

    10) Stop GAB with command :

    /etc/init.d/gab stop

    11) Stop LLT with command :

    /etc/init.d/llt stop

    Press [Enter] to continue:

    Please remember that you should not use "vcsmmconfig -U" and "vxfenconfig -U" for unconfiguring vcsmm and vxfen respectively. For stopping vcsmm and vxfen, please use:

    $ /etc/init.d/vcsmm stop

    $ /etc/init.d/vxfen stop

  13. If the Storage Foundation for Oracle RAC stack does not stop, messages appear checking the status, confirming that Storage Foundation for Oracle RAC is not down and some components are still up. You are prompted to bring the stack down or to reboot. The required commands to bring the stack down are provided again. You are prompted:

    Do you want to continue to bring down SFRAC components? [y,n,q]

    (y)

    You can enter N and safely ignore the step of stopping Storage Foundation for Oracle RAC stack. On selecting N,the installer displays the following message:

    You must manually reboot all the nodes at the same time for the new membership to take effect. To achieve simultaneous reboot of nodes, please shutdown all the nodes at the same time. Now start them only after all are in shutdown state. This step is critical for avoiding those situations where some nodes are running with old membership and the rest with new membership.

    WARNING: If VxFEN has problems coming up after the reboot, SCSI3

    PGR registration keys should be fixed using /opt/VRTSvcs/vxfen/bin/vxfenclearpre command.

    Refer to SF Oracle RAC Installation and configuration guide page #86 for more

    information.

    Press [Enter] to continue:

    This step ensures that all the cluster nodes start with new node IDs on next

    boot. In installsfrac -configure, press Enter to continue. The success of

    the configuration is reported. The configuration summary is saved at:

    /opt/VRTS/install/logs/installsfracxxxxxxxxxx.summary

    The installsfrac log is saved at:

    /opt/VRTS/install/logs/installsfracxxxxxxxxxx.log

  14. If you were able to stop SF Oracle RAC stack successfully, then you do not need to reboot the nodes to start SF Oracle RAC stack. Start the SF Oracle RAC stack on all the nodes using the following steps:

    1) Start LLT with command :

    /etc/init.d/llt start

    2) Start GAB with command :

    /etc/init.d/gab start

    3) Start VXFEN with command :

    /etc/init.d/vxfen start

    4) Start DBAC with commands :

    /etc/init.d/vcsmm start; /etc/init.d/lmx start

    5) Start GMS and ODM with commands :

    /etc/init.d/vxgms start; /etc/init.d/vxodm start

    6) Start vcs with command:

    /etc/init.d/vcs start

    7) Import all previously imported diskgroups and start all volumes.

    Mount all VxFS mounts that were previously mounted.

    8) Start crs and then start database (If any).

  15. Then, using Oracle utilities create Listener and Database appropriately. You may refer to "Creating a starter Database" for database creation.


 ^ Return to Top

< Previous  | TOC | Index | Next  >