Specific Prerequisites

Establishing a Starting Point

In many production environments, you may find that it has been a very long time since the RAC nodes have had the services restarted or have been completely rebooted. This can result in a lack of understanding of the actual state of the RAC cluster and its ability to survive a reboot on its own, prior to installing CTE.

Restarts can uncover issues in the RAC environment that are unrelated to CTE. To avoid issues after a CTE installation, Thales recommends that you restart each RAC node AFTER CTE is installed and PRIOR to establishing any GuardPoints. This may not be feasible in a single node configuration. However, by doing so, CTE is installed but inactive, and you can ensure that the platform is in a workable state prior to getting started.

The Importance of Device Mapping

It is important to use device naming and mapping in a multi-node RAC configuration. Verify the device names to ensure that the disks are mapped to the same disks on each RAC node before applying any GuardPoints. Thales recommends that RAC nodes use the same device names across all nodes. If they do not match, then problems can occur.

If the RAC nodes use the same device names, use a Host Group to create GuardPoints. If they do not match, do not use a Host Group to create GuardPoints. Set them up independently on each Host.

Important Note about Raw Devices on AIX

In general, raw devices are created as either character or block mode devices. Any I/O performed on character devices is non-buffered, while I/O on block devices is buffered and performed in defined block sizes (that is, 4K bytes).

While the Oracle documentation for using ASM with raw devices indicates that you can use either character or block devices, CTE REQUIRES a block device for guarding.

Once guarded, CTE creates both a character and block mode version of the guarded device. Oracle ASM can use either device.