Hitachi

Hitachi Advanced Database Setup and Operation Guide

7.2.3 Operands related to system logs (set format)

[21] adb_log_usrfile_num = number-of-user-log-files

~<integer> ((0 to 10,240)) <<number of real threads that can concurrently update tables>>

Specify the number of user log files required when specific SQL statements and commands are executed.

The number of user log files specified by this value are created when the HADB server is started. For details about the value to specify in this operand, see 6.12.15 Determining the number of user log files.

The following are precautions:

  • If specification of this operand is omitted or 0 is specified, the number of real threads that can update tables concurrently is assumed. As a result, the necessary number of user log files of the size specified in the adb_log_usrfile_size operand is created. This prevents an error from occurring due to a lack of user log files.

  • If you specify 1 in this operand, 2 is automatically assumed. This means that two user log files will be created.

  • Even if you specify for this operand a value that is greater than the number of real threads that can concurrently update tables, the number of real threads that can concurrently update tables is assumed as the value specified for this operand.

  • The generated user log files are not deleted until the database is re-initialized. For this reason, make sure that the space available in the system directory under the DB directory is more than the value calculated by using the formula below.

    Formula

    [Figure]

    Database initialization refers to the execution of the adbinit command when the multi-node function is not used. When using the multi-node function, database initialization refers to carrying out the tasks described in (2) Initializing the file systems subject to node switchover under 16.3.8 Creating a database when creating the database.

Multi-node function

If you are using the multi-node function, there is no need to specify this operand. Even if you specify a valid value, 0 is assumed as the value specified for this operand. In this case, the number of user log files needed by all nodes is automatically calculated and the largest among the calculated values is used.

[22] adb_log_usrfile_size = initial-size-of-user-log-file[,threshold-where-user-log-file-size-is-reduced]

Specify the initial size of the user log file, and the threshold at which a reduction in size of the user log file is triggered.

initial-size-of-user-log-file:

~<integer> ((16 to 1,048,576)) <<16>> (megabytes)

Specify the initial size of user log files in megabytes.

For details about the value to specify in this operand, see 6.12.2 Determining the size of the user log files.

Multi-node function

When using the multi-node function, the largest value specified for initial-size-of-user-log-file among all nodes takes effect.

threshold-where-user-log-file-size-is-reduced:

~<integer> ((0, 32 to 2,097,152)) <<initial-size-of-user-log-file x 2>> (megabytes)

Specify in megabytes the threshold at which the size of the user log file is to be reduced. If the size of the user log file used by an update transaction exceeds the size specified for this operand when the update transaction is settled, the user log file is reduced to its initial size. Normally, there is no need to specify this operand.

The rules are as follows:

  • If you specify 0 for threshold-where-user-log-file-size-is-reduced, reduction in size of the user log file will not take place.

  • If threshold-where-user-log-file-size-is-reduced < initial-size-of-user-log-file, then the default value (initial-size-of-user-log-file x 2) is assumed.

Multi-node function

  • When using the multi-node function, the largest value specified for threshold-where-user-log-file-size-is-reduced among all nodes takes effect.

  • If the value of threshold-where-user-log-file-size-is-reduced that takes effect is less than the value of initial-size-of-user-log-file that takes effect, then a value equivalent to twice the initial size of the user log file is assumed for threshold-where-user-log-file-size-is-reduced.

  • If you specify 0 for threshold-where-user-log-file-size-is-reduced on all nodes, the size of the user log file will not be reduced.

[23] adb_log_usrbuf_num = number-of-user-log-buffers

~<integer> ((64 to 12,582,912)) <<64>>

Specify the number of buffers used for accessing user log files.

Normally, there is no need to specify this operand. Increasing the value might improve the speed of accessing user log files in some cases.

Note that the HADB server allocates a user log buffer to each user log file.

Multi-node function

If you are using the multi-node function, the largest among the numbers of buffers used for accessing user log files specified on the individual nodes is used.

[24] adb_log_rec_msg_interval = output-interval-of-message-showing-database-recovery-processing-time-elapsed

~<integer> ((0 to 60)) <<5>> (minutes)

Specify the interval at which a message is output indicating how much time (in minutes) has elapsed when performing database recovery processing.

Normally, there is no need to specify this operand.

During database recovery processing, the following two messages are output according to the value specified for this operand:

■ Cases in which the KFAA81215-I message is output

When database recovery processing is initiated by processing such as the following, a message KFAA81215-I indicating how long database recovery processing has taken so far is output at a fixed interval determined by the value of this operand.

  • Restarting the HADB server

  • Rolling back a transaction

For example, if 3 is specified for this operand, the message KFAA81215-I is output every three minutes while database recovery processing is in progress. You can use the contents of these KFAA81215-I messages to follow the progress of database recovery processing.

■ Cases in which the KFAA81211-I message is output

The message KFAA81211-I is also output during database recovery processing when the recovery processing for a particular transaction is taking a long time.

For example, if 3 is specified for this operand, the message KFAA81211-I is output when a single transaction has taken three minutes or longer to recover, and when recovery of that transaction has completed. The KFAA81211-I message contains information related to the recovery processing for the transaction in question. You can use this output information as part of the resources you use to identify the cause of a delay in transaction recovery processing.

If you specify 0 for this operand, neither message KFAA81211-I nor KFAA81215-I is output.

To Page Top

All Rights Reserved. Copyright (C) 2012, 2020, Hitachi, Ltd.