20.3.2 Creating DB directories (when using the multi-node function)

When the multi-node function is being used, some of the DB areas are shared among all HADB servers in the multi-node configuration. For details about the multi-node function, see the topic Multi-node function in Architecture in the HADB Setup and Operation Guide. For details about the DB areas that are shared among all HADB servers in a multi-node configuration, see the topic DB area file configuration in DB areas (areas for storing tables and indexes) in Architecture in the HADB Setup and Operation Guide.

To share DB areas, the different systems must be able to uniquely identify a device allocated to DB area files. For this purpose, WWN is used. WWN is a fixed identifier that does not depend on the system and that is allocated to each device on the HBA.

If you will be using the multi-node function, when you create the DB directory, allocate to each DB area to be shared a block special file that includes WWN (path beginning with /dev/disk/by-id/wwn-).

When the multi-node function is being used, execute the adbinit command using the following procedure to create the DB directory.

Procedure

Create an initialization option file for the master node.
Execute the adbinit command on the master node.
Create an initialization option file for each slave node.
Execute the adbinit command on each slave node.

Perform steps 3 and 4 for each slave node on each HADB server.

Note that you can also specify block special files for work table DB areas in the server definition. For details, see the description of the adb_blk_path_wrk operand in Operands related to system configuration (set format) in Detailed descriptions of the server definition operands in Designing the Server Definition in the HADB Setup and Operation Guide.

Using an example, the following subsections explain each of these steps.

Organization of this subsection

(1) Creating an initialization option file for the master node
(2) Executing the adbinit command on the master node
(3) Creating an initialization option file for each slave node
(4) Executing the adbinit command on each slave node

(1) Creating an initialization option file for the master node

Create an initialization option file and specify the initialization options as follows:

Specify Y in the adb_init_dbarea_initialize operand because each DB area for the master node must be initialized.
Allocate a block special file that includes a WWN identifier to each and every DB area that will be shared among all HADB servers in the multi-node configuration.
If you want to allocate a block special file to a DB area (work table DB area) that will not be shared among the HADB servers in the multi-node configuration, allocate a block special file specific to the master node.

Example of an initialization option file for master node

set adb_init_dbarea_initialize = Y
set adb_init_wrk_blk_path = /dev/mapper/adb-ADBWRK_lv
set adb_init_mst_blk_path = /dev/disk/by-id/wwn-0x600605b003a2e280172ab596258093bc
set adb_init_dic_blk_path = /dev/disk/by-id/wwn-0x600605b003a2e280172ab5962580e259
set adb_init_stbl_blk_path = /dev/disk/by-id/wwn-0x600605b003a2e280172ab5962581311b
adbinitdbarea -n DBAREA1 -i 1G,2 -v \
  /dev/disk/by-id/wwn-0x60060e80102172300511166300000000,\
  /dev/disk/by-id/wwn-0x60060e80102172300511166300000001

To Page Top

(2) Executing the adbinit command on the master node

On the master node, execute the adbinit command specifying the name of the initialization option file created in (1) Creating an initialization option file for the master node.

For details, see (2) Executing the adbinit command in 20.3.1 Creating DB area files using block special files.

To Page Top

(3) Creating an initialization option file for each slave node

Copy the initialization option file created in (1) Creating an initialization option file for the master node to each slave node, and then change the initialization option settings as follows:

Specify N in the adb_init_dbarea_initialize operand.
If a block special file specified in the initialization option file for the master node satisfies either of the following conditions, change the block special file name or delete the corresponding specification:
- A block special file is specified for a DB area that will not be shared among the HADB servers in the multi-node configuration.
- There is no block special file that is specific to the slave node and has the same name as a block special file specified for a DB area that will not be shared among the HADB servers in the multi-node configuration.
Do not change the settings for the other operands.

Example of an initialization option file for a slave node

set adb_init_dbarea_initialize = N
set adb_init_mst_blk_path = /dev/disk/by-id/wwn-0x600605b003a2e280172ab596258093bc
   :

To Page Top

(4) Executing the adbinit command on each slave node

On the slave node, execute the adbinit command specifying the name of the initialization option file created in (3) Creating an initialization option file for each slave node.

For details, see (2) Executing the adbinit command in 20.3.1 Creating DB area files using block special files.

To Page Top