System log files

Question

When HiRDB starts normally, doesn't it continue to use the system log file that was being used during the previous HiRDB session?

Case 1

I executed normal startup or forced normal startup of HiRDB, but the KFPS01224-I message was displayed indicating that there was no system log file available for swapping. When this problem occurred, the system log files were in the following status:

There was adequate free space remaining in the system log file that was being used during the previous HiRDB session.
There was an available system log file that could be swapped in.

Case 2

I executed normal startup or forced normal startup of HiRDB, but the KFPS01256-E message was displayed indicating that there was no available system log file, and then HiRDB (or unit for a HiRDB parallel server configuration) terminated abnormally with abort code Psjnf07. When this problem occurred, the system log files were in the following status:

There was adequate free space remaining in the system log file that was being used during the previous HiRDB session.
There was no available system log file that could be swapped in.

Answer: When HiRDB starts normally, it does not continue to use the system log file that was being used during the previous HiRDB session. Because HiRDB swaps system log files during a normal startup (including a forced normal startup), the current file is swapped. The KFPS01256-E warning message was displayed because the last system log file available to be swapped in was allocated, and now there are no more system log files available for future swapping.; On the other hand, when HiRDB is restarted, the system log files are used as if the previous session was being resumed (system log files are not swapped).; If the same system log files were used after a normal startup, the system log from the previous session and the system log for the current session would be stored in a single system log file. In this case, performing file management tasks (such as unloading into unload log files) would be complicated. Therefore, HiRDB swaps system log files during normal startup so that a system log file containing only the new information following the normal startup is compiled.; When pd_log_rerun_swap=Y is specified, the system log files are also swapped during a restart. If a shortage of space occurs in the system log file, the restart fails; therefore, the number of system log files available to be swapped in (in swappable target status) should be increased.

(2) The KFPS01251-E message was displayed indicating that a system log file was invalid

Question: I executed forced normal startup of HiRDB (or a unit) after forced termination occurred (or the HiRDB unit terminated abnormally due to a software or hardware error), but the KFPS01251-E message was displayed indicating an invalid system log file. As a result, the system log file could no longer be used as a swappable target. What is the cause of this and what action should be taken?

Answer

Cause:: The file is an incomplete file because the primary file was not closed normally when HiRDB (or the unit) terminated abnormally.

Action:: Use the pdlogunld command to unload the system log in this system log file. If this system log information is not needed, the pdlogchg command can be used to simply change the file's status.

(3) Addition of a system log file and restart of HiRDB were repeated several times

Question: HiRDB terminated abnormally immediately after a restart because there was no system log file available to be swapped in (in swappable target status). So I allocated a new system log file and restarted HiRDB. HiRDB terminated abnormally again due to a shortage of system log files. I repeated the process of adding a system log file and restarting HiRDB, but the problem could not be resolved. Why did HiRDB keep terminating abnormally?

Answer

Possible reasons:

pd_log_rerun_swap=Y is specified.
When Y is specified in this operand, the current system log files are swapped even during a restart, thus reducing the number of swappable files by one. Therefore, this event cannot be eliminated by adding only one system log file. Add at least two system log files and restart HiRDB.
There is not enough space in the overwrite-enabled system log file.
During a HiRDB restart, system log information is collected even for rollback. Therefore, repeating the restart increases the number of system log files that are output, thereby increasing the range of the overwrite-disabled system log files. Because a new synchronization point dump is collected upon completion of all rollbacks after a restart, if the unit terminates abnormally after this synchronization point, the range of the overwrite-disabled system log files will not increase.
At least two system log files should be added at the same time, rather than adding one at a time. The total size of the files to be added should be at least 1.5 times the system log input range at the first restart when this event occurred for the first time (from the location indicated in the KFPS01262-I message to the location indicated in the KFPS01263-I message).

Remarks: This problem can be prevented by providing multiple reserved file groups and specifying pd_log_rerun_reserved_file_open=Y.

Appendix A.1 System log files

(1) The KFPS01224-I warning message was displayed indicating that there are no system log files in swappable target status

(2) The KFPS01251-E message was displayed indicating that a system log file was invalid

(3) Addition of a system log file and restart of HiRDB were repeated several times