LVM Hierarchy Creation Procedures

To create a hierarchy in which a file system or higher-level application uses an LVM logical volume, the following high-level procedure should be followed.

1. Determine the desired configuration of your LVM volume groups and logical volumes. In doing this, keep in mind the following points:

• All of the disk resources associated with a given volume group must move together from one server to another in the SPS cluster.

• All of the logical volumes associated with a given volume group (and any file systems or applications which use them) must move together from one server to another in the SPS cluster.

• (SLES and RHEL7 only) lvm2-lvmetad is disabled during the LVM RK installation. If you install lvm2 after installing the LVM RK, you need to disable it manually. Refer to the operating system documentation for more information.

2. On the system which is to be the primary server for your application, create and activate the desired volume groups and logical volumes using the tools provided by the LVM package and described in the LVM HowTo document referenced in the Documentation and References topic.

If you are using shared storage, you must ensure that all physical volumes assigned to a volume group are properly shared between the machines in the SPS cluster on which you intend to run the protected application. If you intend to use LVM with DataKeeper, see the Using LVM with DataKeeper topic.

3. Create file systems on each of the logical volumes. If instead you intend to use raw I/O, bind a raw device to each of the logical volume devices.

4. Configure the protected application on the file systems following the configuration instructions in the administration guide for the SPS recovery kit associated with the application.

Create and extend the application hierarchy following the instructions in the appropriate application recovery kit administration guide.

!IMPORTANT: Perform manual in-service operations to temporarily move the application hierarchy to each of the cluster nodes to which the hierarchy has been extended. This step must be done once prior to any node failover operations in order for the LVM subsystem on each cluster node to know about the configuration of the new volume groups and logical volumes. After you have performed these manual switchovers, move the application hierarchy back to the desired primary cluster node.