Nonstop Database, HiRDB Version 9 System Operation Guide

[Contents][Index][Back][Next]

3.8.2 Environment setup

To use the automatic log unloading facility, the following environment settings are required:

Organization of this subsection
(1) Creation of directories for unload log files
(2) Specification of the pd_log_auto_unload_path operand
(3) Specification of the pd_log_auto_unload_restart operand (for retrying if an error occurs)

(1) Creation of directories for unload log files

The HiRDB administrator creates directories for unload log files. It is wise to check the availability of disk space before creating a directory. If the amount of available disk space is small, creation of the directory might result in a disk space shortage.

Assign write privileges, read privileges, and execution privileges to the HiRDB administrator for the created directory.

A HiRDB file system area for utilities must be created in order to create a directory for unload log files in a HiRDB file system area.

In the case of a HiRDB parallel server configuration, create a different directory for each server.

(a) Required disk space

The required disk space can be determined from the following formula:

Required disk space (bytes) =
(total-number-of-records-in-system-log-files-to-be-unloaded#1 [Figure]
system-log-file-record-length)#2 [Figure]
number-of-unload-log-files-to-be-created-in-directories [Figure] 1.2

Notes
  • Note that the disk space requirement computed here is not for a single directory; it is the total disk space requirement for all directories if multiple directories are created.
  • When the disk space available for a single directory is not adequate, create multiple directories.

#1: If the system log file automatic extension facility is applied, calculate the number by using the value specified in the upper extension limit of the pd_log_auto_expand_size operand.

#2: This is an estimated value for the system log file.
(b) Number of directories for unload log files

Although 128 directories for unload log files can be created, creating too many is not advisable. Four directories at the most are recommended so that their management does not become cumbersome. While a single directory is easy to manage, the automatic log unloading facility will stop if the disk becomes full or an error occurs.

When multiple directories are available, the automatic log unloading facility does not stop as long as unload log files can be created in at least one of the directories. For this reason, create the directories in different partitions.

(c) Important
(d) Creating unload log files in a HiRDB file system area

To create unload log files in a HiRDB file system area, a HiRDB file system area must be created with the pdfmkfs command. The following table provides guidelines for the options to be specified in the pdfmkfs command.

Table 3-8 Guidelines for options to be specified in the pdfmkfs command (when unload log files are to be created in a HiRDB file system area)

pdfmkfs command option Specification guideline
-k For the purpose, specify UTL.
-n For the HiRDB file system area length, specify the value determined by the following formula:
total-number-of-records-in-system-log-files-to-be-unloaded [Figure]
system-log-file-record-length [Figure]
number-of-unload-log-files-to-be-created [Figure] 1.2 [Figure] 1,048,576
-l For the maximum files count, specify the number of unload log files to be created.
-e For the maximum increments count, specify the number of unload log files to be created [Figure] 24.

(2) Specification of the pd_log_auto_unload_path operand

Specify in the pd_log_auto_unload_path operand the directories that have been created for unload log files.

When unload log files are to be created in a HiRDB file system area, specify the name of the HiRDB file system area.

(3) Specification of the pd_log_auto_unload_restart operand (for retrying if an error occurs)

If you specify Y for the pd_log_auto_unload_restart operand or omit the specification, and if unloading fails due to a temporary error, it will be retried during the next system log file swap.