2.3. Configuring Pools

Configuring pools (GFS volumes) consists of creating pool configuration files for the GFS Cluster Configuration Archive (CCA) and the Oracle files, creating disk partitions for the pools, creating each pool (running the pool_tool -c for each pool), and activating the pools. For reliability, striping across multiple RAID arrays is preferred to single drives.

	Note
	The CCA is the repository for the GFS Cluster Configuration System (CCS) configuration files. For more information about CCS and CCA, refer to the Red Hat GFS Administrator's Guide for GFS 6.0.

Example configuration files provided in this procedure are based on Table 1-4, and can be adapted to the hardware environment on which your GFS is being deployed.

To configure the pools, follow these steps:

1. Create and save pool configuration files for the Oracle cluster and server control software (this supports the cluster manager and is not used for the database). Suggested names for the pools are oraraw1 and oraraw2, with corresponding pool configuration files, oraraw1.pool and oraraw2.pool (refer to Example 2-2 and Example 2-3). For the sample cluster in this manual, the size of each of these pools is 100 MB.

   poolname oraraw1 subpools 1 subpool 0 0 1 gfs_data pooldevice 0 0 /dev/sda1

   Example 2-2. oraraw1.pool

   poolname oraraw2 subpools 1 subpool 0 0 1 gfs_data pooldevice 0 0 /dev/sda2

   Example 2-3. oraraw2.pool

2. Create and save a pool configuration file for the GFS CCA. The suggested name for the pool is ccs_archive, with the corresponding pool configuration file, ccs_archive.pool (refer to Example 2-4). The recommended size of this pool is relatively small — for example, 8 MB. For the sample cluster in this manual, the size of this pool is 10 MB.

   poolname ccs_archive subpools 1 subpool 0 0 1 gfs_data pooldevice 0 0 /dev/sda3

   Example 2-4. ccs_archive.pool

3. Create and save a pool configuration file for the $ORACLE_HOME directory. The suggested name for the pool is oracle_base, with the corresponding pool configuration file, oracle_base.pool (refer to Example 2-5). The recommended size of this pool is at least 10 GB. For the sample cluster in this manual, the size of this pool is 20 GB.

   poolname oracle_base subpools 1 subpool 0 0 1 gfs_data pooldevice 0 0 /dev/sda4

   Example 2-5. oracle_base.pool

4. Create pool configuration files for the undo and index tablespace data files. Suggested names for the pools are oraundo and oraindex, with the corresponding pool configuration files, oraundo.pool and oraindex.pool (refer to Example 2-6 and Example 2-7). The recommended size of each pool is greater than 10 GB. For the sample cluster in this manual, the size of each of these pools is 98 GB.

   poolname oraundo subpools 1 subpool 0 0 1 gfs_data pooldevice 0 0 /dev/sdb1

   Example 2-6. oraundo.pool

   poolname oraindex subpools 1 subpool 0 0 1 gfs_data pooldevice 0 0 /dev/sdc1

   Example 2-7. oraindex.pool

5. Create and save a pool configuration file for Oracle system and data files. The suggested name for the pool is oradata, with the corresponding pool configuration file, oradata.pool (refer to Example 2-8). For the sample cluster in this manual, the size of this pool is 60 GB.

   poolname oradata subpools 1 subpool 0 0 1 gfs_data pooldevice 0 0 /dev/sdd1

   Example 2-8. oradata.pool

6. Create disk partitions according to Table 1-4. For example, start creating the disk partitions as follows:

   # fdisk /dev/sda

   # fdisk /dev/sdb

   # fdisk /dev/sdc

   # fdisk /dev/sdd

7. Create the required GFS pools using the GFS pool_tool -c command. You just need to do this from one node in the cluster. Refer to Example 2-9.

   rac1 # pool_tool -c oraraw1.pool Pool label written successfully from oraraw1.pool. rac1 # pool_tool -c oraraw2.pool Pool label written successfully from oraraw2.pool. rac1 # pool_tool -c ccs_archive.pool Pool label written successfully from ccs_archive.pool. rac1 # pool_tool -c oracle_base.pool Pool label written successfully from oracle_base.pool. rac1 # pool_tool -c oraundo.pool Pool label written successfully from oraundo.pool. rac1 # pool_tool -c oraindex.pool Pool label written successfully from oraindex.pool. rac1 # pool_tool -c oradata.pool Pool label written successfully from oradata.pool.

   Example 2-9. Creating GFS Pools

8. Allow your systems to see the pools that were created in the previous step by running the pool_assemble command on all nodes. Example 2-10 shows running the pool_assemble command on node rac1:

   rac1 # pool_assemble ccs_archive assembled. oracle_base assembled. oradata assembled. oraindex assembled. oraraw1 assembled. oraraw2 assembled. oraundo assembled.

   Example 2-10. Assembling Pools

9. Proceed to Section 2.4 Setting Up the GFS Configuration.