You can assign any name to the HiRDB directory, but be sure to specify all the following information. The pathname of the HiRDB directory must be no longer than 128 characters (bytes). For Linux, specify 1-27 characters (bytes) as the pathname.

• Directory name: Any name
• Owner: HiRDB administrator
• Group: HiRDB group
• Mode: 0755

Also note the following:

• Specify a character string that begins with / (forward slash) and that consists of only the following characters:
  
  • Alphanumeric characters
  • _ (underscore)
  • . (period)
  • / (forward slash separating paths)
• The forward slash (/) alone cannot be specified.
• A path name must not exceed 128 characters (bytes). For the Linux version, a path name must not exceed 118 characters (bytes).

1. To create the HiRDB directory, you need at least 2 gigabytes of disk space. This is just a guideline; the actual size that is required depends on the size of the system.
2. You should not create the HiRDB directory in the root directory. Files containing troubleshooting information are created in the HiRDB directory (files under $PDDIR/spool). If you do not delete these files periodically using the pdcspool command, a space shortage may occur on the disk, resulting in adverse effects on operating system operation.
   If you do create the HiRDB directory in the root directory, you should use a different partition than that of the root directory.
   
3. Create the HiRDB directory on a local disk. Do not specify nosuid as the mount type of the file system in /etc/checklist or etc/fstab.
4. Do not use the HiRDB directory as the target of a symbolic link.
5. For multi-HiRDB, create a HiRDB directory with a different name for each HiRDB.
6. To employ a 2-to-1 or mutual system switchover organization for a HiRDB parallel server configuration, you cannot share the same HiRDB directory among all server machines. For details about how to define the HiRDB directory, see Example of mutual switchover configuration in System configuration examples of a HiRDB parallel server configuration in the HiRDB Version 9 System Operation Guide.

HiRDB outputs troubleshooting information to $PDDIR/spool in situations such as when a server process or client is forcibly terminated. When a command or utility is aborted by pressing Ctrl + C (or some other method) and no output destination has been explicitly specified for work files, temporary work files output under $PDDIR/tmp by the command or utility are retained rather than deleted. These files are often the cause of a shortage of disk space available to the HiRDB directory. Because insufficient disk space for the HiRDB directory can cause abnormal termination, HiRDB deletes the following files periodically:

• Troubleshooting information files (files in $PDDIR/spool)
• Temporary work files (files in $PDDIR/tmp)
• Files in the directory specified in the pd_tmp_directory operand

For details about these consistently increasing files, see 1.3.2 Files that consistently increase in size.

Normally, these files are deleted every 24 hours. The interval between deletions can be changed with the pd_spool_cleanup_interval operand. You can also specify deletion of only those files output prior to the date specified by the pd_spool_cleanup_interval_level operand.

It is also possible to delete all at once all troubleshooting information (all files in $PDDIR/spool).

• Use the pdcspool command to delete the troubleshooting information files. Temporary work files (in $PDDIR/tmp) can also be deleted.
• Automatically delete the troubleshooting information files during the HiRDB startup. In this case, use the pd_spool_cleanup operand to specify whether to delete the troubleshooting information files. If you omit this operand, the troubleshooting information files are deleted automatically.
  You can also specify deletion of only those troubleshooting information files output prior to the date specified by the pd_spool_cleanup_level operand.
  

Note

You can select the troubleshooting information to be deleted using the pdcspool command's option, pd_spool_cleanup_level operand or pd_spool_cleanup_interval operand.

To prepare for possible errors on the disk that contains the HiRDB directory, make a backup copy of the files in the HiRDB directory (files under $PDDIR/conf). You need the backup copy of the files under $PDDIR/conf to restore the HiRDB directory. There are HiRDB system definition files under $PDDIR/conf. If you have modified the HiRDB system definition, back up the files contained under $PDDIR/conf.

If you have created user files under the HiRDB directory, also back up those files. You need this backup copy to restore the HiRDB directory.

For details about how to recover the HiRDB directory, see the HiRDB Version 9 System Operation Guide.

• Backing up the disk that contains the HiRDB directory
  If you need to back up the disk that contains the HiRDB directory, use the following procedure:
  
  1. Terminate HiRDB with the pdstop command.
  2. Execute the pdsetup -d command. Enter n as a response.^#
  3. Use the OS function (command) to back up the disk that contains the HiRDB directory.
  4. Execute the pdsetup command.
  5. Enter the pdstart command to start HiRDB.

  #: You need to execute the pdsetup -d command to remove all the files that may cause problems after recovery. Additionally, you need to enter n as a response to retain all the files that are required after recovery.