OpenTP1 Version 7 Operation

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

3.13.1 Handling resources

To run OpenTP1 without stopping, you need to keep the number of resources to be used from increasing. OpenTP1 controls resources as follows:

Serial numbers
If the serial number of journal blocks and other items reaches the maximum, it is wrapped around to 0 or 1. Therefore, there is no problem in nonstop operation.

Memory requirements
The memory requirements do not increase during nonstop operation.

Disk space
The following table shows how each file uses disk space.

Table 3-30 Disk space used by OpenTP1

Directory Description
/tmp Stores the standard output and standard error output file (betran.log) for the processes under OpenTP1. This file may increase unlimitedly and therefore we recommend that you create a mechanism for deleting the files when the machine starts up, or to perform generation management of betran.log. See 3.5 Standard output file operations.
$DCDIR/bin The required disk space does not increase.
$DCDIR/lib The required disk space does not increase.
$DCDIR/include The required disk space does not increase.
$DCDIR/etc The required disk space does not increase.
$DCDIR/aplib The required disk space does not increase.
$DCDIR/tmp Used as the home directory of every process. If a process goes down, a core file is created in this directory and the file is moved to $DCDIR/spool/save. Therefore, the required disk space increases temporarily but it will soon return to the original size. The directory is cleared when online operations resume.
$DCDIR/spool/save/* A core file is saved. Since only three generations of the core file are created, the required disk space does exceed the limit.
$DCDIR/spool/dclckinf/* This directory stores deadlock information and timeout information. You should delete the data at regular intervals#1.
$DCDIR/spool/dctrninf/* This directory stores the undecided-transaction information when you specify trn_tran_recovery_list=Y in the transaction service definition.#2
Although $DCDIR/spool/dctrninf/trace stores trace data for the XA interface, a wraparound occurs when its size reaches the preset limit. Therefore, the required disk space does not exceed the limit.
$DCDIR/spool/errlog/* This directory stores OpenTP1 error log information. Since a wraparound occurs when the size of the information reaches the preset limit, the required disk space does not exceed the limit.
$DCDIR/spool/cmdlog/* This directory stores the OpenTP1 command log. Since a wraparound occurs when the size of the log reaches the preset limit, the required disk space does not exceed the limit.
$DCDIR/spool/dcjnlinf/* This directory stores invalid journal block information. No file is usually created in this directory.
$DCDIR/spool/dcjnlinf/unload/* This directory stores the unload journal files generated by the automatic unloading facility when jnl_auto_unload=Y is specified in the system journal service definition and no value is specified for the jnl_auto_unload_path operand. We recommend that you delete the files you no longer need at regular intervals.
$DCDIR/spool/dctjlinf/* $DCDIR/spool/dcsjl/* Do not delete the files created in these directories.
$DCDIR/spool/trnrmcmd/* This directory stores the object file created by the trnmkobj command.
$DCDIR/spool/olkfifs/* $DCDIR/spool/olkrsfs/* The file size is fixed and the number of files corresponds to the definition. Therefore, the number of files does not exceed the limit.
$DCDIR/spool/dcprfinf/* The file size and the number of files are fixed. In addition, a wraparound occurs when its size reaches the preset limit. Therefore, the required disk space does not exceed the limit.
$DCDIR/spool/ This directory stores traces such as dclog1 and dclog2. Since a wraparound occurs with the OpenTP1 trace when its size reaches the preset limit, the required disk space does not exceed the limit. Shared memory dumps are also output to this directory. Since the size of the dump depends on the definition, the required disk space does not exceed the limit.

#1
When you specify lck_deadlock_info_remove_level, you can automatically delete the information at regular intervals.

#2
Since the information accumulates at each rerun, delete it at regular intervals. When you specify trn_recovery_list_remove_level, you can automatically delete the information at regular intervals.