Surviving the Reboot and Failover Testing
Preparing for Failover Testing with ASMLib
When using ASMLib with the createdisk command, there is no requirement to make additional changes in rc.local or other areas for mapping device names or to use chmod or chown for SecVM. This is because it is managed for you by the createdisk function and you can verify this by running the following command:
ls -l /dev/oracleasm/disks
CTE Load Order and Startup Scripts
The last change is to ensure that CTE starts before ASM starts in the startup scripts.
Failover Testing
Confirm that everything is functional:
-
Ensure that the GuardPoints are all operational.
-
Ensure that you receive valid results when you query the database.
-
Verify that the load order ensures that CTE starts before ASM.
Once verified, you can start the failover testing for each RAC Node.
-
Reboot the RAC Node 1 and monitor the startup.
-
Once the restart is clean, reboot RAC Node 2 and monitor the startup.
Issues with Device Mapper and Invalid Guard Path
If CTE is unable to apply a GuardPoint on a raw device, the logs may generate an error similar to the following:
[SecFS, 0] EVENT: Failed to guard /dev/mapper/devicename (reason: Invalid Guard Path flags 0x2 gypped 0x4 status 0x11) – Will retry later
If you receive this error, use the setup command to check the status of the disks, type:
setup info <deviceName>
Before attempting to establish a GuardPoint, look closely at the open count value and ensure that it is 0 on all nodes.
