Procedure to prepare SSD after a disk replacement on IBM Smart Analytics Systems

Content

Procedure to prepare SSD after a disk replacement on IBM Smart Analytics System

Assumptions

• The database is operational on the primary or standby node(s).
• Users are connected to the database and operational.
• One or more SSD has been replaced with a new drive.
• All database partitions have the same temporary spaces and sizes defined.

Overview
The following is typically the quickest method. The example used in this section shows how to make the directories and copy the appropriate files to support the default temporary space delivered with the system. If you have added additional temporary spaces, you should follow the same procedure replacing the names accordingly to ensure all tablespaces have the correct supporting disk structure and file available on the replaced SSD.

Naming conventions
This example uses the following naming conventions:

This example creates the directory structure and file supporting a temporary tablespace created with the following command.
Example of the command used to create a temporary tablespace.

You may have more than one temporary tablespace. Repeat this process for all tablespaces to ensure all the appropriate files are copied to the new SSD.

Procedure
1. Login as root user to the server where the SSD was replaced.

2. After replacing the SSD and reboot, there should be 8 ssd drives mounted on the data nodes.

Validate: Use df or mount commands to validate the drives are mounted.

For example

Command: df -g | grep ssd

Example output

3. If any SSD is not mounted, use the mount command to mount it

Command: mount /db2ssd/bcuaix/ssd_unmounted_drive

Example: mount /db2ssd/bcuaix/ssd3

Validate all drives are mounted (see step #2 above) before moving to the next step.

4. Make directory

Command: mkdir -p /db2ssd/bcuaix/ssd_replaced_drive/dbname

Example: mkdir -p /db2ssd/bcuaix/ssd7/BCUDB

5. Change ownership of new directory to match the existing directory

Command: chown instance_owner:instance_group /db2ssd/bcuaix/ssd_replaced_drive/dbname

Example: chown bcuaix:bcuigrp /db2ssd/bcuaix/ssd7/BCUDB

6. Change the permission of the new directory to match the existing directory

Command: chmod 711 /db2ssd/bcuaix/ssd_replaced_drive/dbname

Example: chmod 711 /db2ssd/bcuaix/ssd7/BCUDB

7. Copy file(s) or ftp from a similar server
NOTE: the file(s) will be large. It is quickest to copy the file from a neighboring SSD. If a neighboring SSD is not available or has a different structure, you can alternatively compress and ftp the file from a different node in the cluster that has the same temporary space name and size.

Command: cp /db2ssd/bcuaix/ssd_existing_drive/dbname/temp16k /db2ssd/bcuaix/ssd_replaced_drive/dbname/

Example: cp /db2ssd/bcuaix/ssd0/BCUDB/temp16k /db2ssd/bcuaix/ssd7/BCUDB/

8. Change ownership of new file to match the existing file

Command: chown instance_owner:instance_group /db2ssd/bcuaix/ssd_replaced_drive/dbname/temp_space_name

Example: chown bcuaix:bcuigrp /db2ssd/bcuaix/ssd7/BCUDB/temp16k

9. Change the permission of the new file to match the existing file

Command: chmod 600 /db2ssd/bcuaix/ssd_replaced_drive/dbname/temp16k

Example: chmod 600 /db2ssd/bcuaix/ssd7/BCUDB/temp16k

NOTE: After completing this step for all temporary tablespaces, the server is prepared to start DB2 resources on this server.

10. Validate the permission, owner, group, and appropriate files exist.

Command 1: ls /db2ssd/bcuaix/ssd_replaced_drive/
Command 2: ls /db2ssd/bcuaix/ssd_replaced_drive/dbname

Example Output
The following output displays the correct name, permission, owner, and group

11. Move resources as appropriate.
If the machine that has failed SSD was a primary server, the resources failed over to the standby. When you can schedule an outage window, you will need to move resources back to the primary. If the failures were on the standby node, your task is complete.

To move resources from the standby to the primary, run the following command.

Command: hafailback primary_server_name

Example: hafailback bcudata01

12. Use hals or lssam to monitor the resources as the move from the standby to the primary server. The move may take a few minutes. After the successful move, resources will show Online on the primary server.