2.7.1 Embedded database processing when a problem occurs and the action to be taken by the embedded database system administrator
This subsection describes the following items related to the processing of the embedded database when a problem occurs, and the action to be taken by the embedded database system administrator.
- Organization of this subsection
(1) The responsibilities of the embedded database system administrator when a problem occurs
When a problem occurs, the embedded database system administrator collects the troubleshooting information output to the following locations:
- In Windows:
-
-
Files under embedded-database-practical-directory\spool
-
Files under embedded-database-practical- directory\tmp
-
Files under embedded-database-practical- directory\uxpldir
-
Windows event log
-
- In UNIX:
-
-
Files under embedded-database-practical-directory/spool
-
Files under embedded-database-practical-directory/tmp
-
syslog
-
- Remarks
-
The information required to troubleshoot a problem is described in 2.7.7 Information to be collected. The embedded database system administrator must refer to this information when troubleshooting.
(2) The information obtained by the embedded database when a problem occurs
The following table describes the troubleshooting information obtained by the embedded database when a problem occurs.
|
Information |
Description |
|---|---|
|
Message log files
|
Messages output by the embedded database. Back up this information when a problem occurs. |
|
Standard output and standard error output for command execution |
Information and error messages output by commands. This information appears on the screen from which the commands were entered. To save the information to a file, redirect the standard output and standard error output as needed. |
|
Messages output by the embedded database. In UNIX, use an OS editor to view this information. Note that messages might not be output when too many processes are attempting to access syslog. |
|
Core save files# (in UNIX) embedded-database-practical-directory/spool/save/file-name |
Data and stack information obtained in UNIX systems from processes related to the embedded database. Because only three such files can be saved at any one time, back up the core save files that you want to keep. The file name format is ajs2n, where n is the sequence number of the core save file (1 to 3). Note that in some cases, sequence numbers might not be assigned. |
|
Abort information file#
|
Abort information. Back up this information if it is output. If the abort information consists of abort codes only, you can view them by using any text editor. The file name format is abcode.process-ID-of-server-process. |
|
Snap information captured at the time the problem occurred#
|
Snap information captured at the time the problem occurred. Back up this information if it is output. The file name format is ajs2n.deb, where n is the sequence number of the file (1 to 3). Note that in some cases, sequence numbers might not be assigned. |
|
Shared memory dump file#
|
Data collected from shared memory by the embedded database. Back up this information if it is output. The file name format is ajs2.rmb.process-ID-of-server-process. |
|
Simple dump file#
|
Data collected from shared memory and process-specific memory by the embedded database. Back up this information if it is output. The file name is the combination of the date and the process ID. |
|
Command trace files
|
The history information of executed commands, including commands that the embedded database generates internally. Back up the files before viewing this information in any text editor. |
|
Error log files
|
Internal information output by the embedded database. Back up this information if it is output. |
|
Connected-user information file
|
Information about users who were connected when the embedded database terminated. Use any text editor to view this information. |
|
Connected-user details file
|
|
|
Locked resource management table information file
|
User information when a deadlock, lock-wait timeout, or insufficient locked-resource management table error occurs due to exclusive control by the embedded database. Use any text editor to view this information. Back up this information if it is output. |
- #
-
This troubleshooting information is output whenever the server process of the embedded database terminates abnormally. You can tell whether this process has terminated abnormally by checking for output of the KFPS01820-E message. If the message has been output, you can check the server name, process ID, and end state. Note that the following circumstances under which troubleshooting information might not be output.
-
If the end state begins with the letter c or d, no troubleshooting information will be output.
-
If the end state is 007f in Windows or 0009 in UNIX, the abort information file, core save files, and shared memory dump file will not be output.
-
(3) Embedded database processing when a problem occurs
The following describes the processing performed by the embedded database performs when a problem occurs.
(a) System recovery when the embedded database terminates abnormally
If, after the embedded database has terminated abnormally, the cause of the problem is eliminated and the embedded database is restarted, the embedded database restores the system to the status existing before the problem occurred.
(b) Processing when the embedded database repeatedly terminates abnormally
If the embedded database terminates abnormally, restart processing is executed. However, if the restart processing terminates abnormally three times in succession within 30 minutes, no more attempts to restart will be made. At this point, the embedded database system administrator must eliminate the cause of the problem, and then use the ajsembdbstart command to restart the embedded database.
(4) Action to be taken when a problem occurs in an embedded database process
When a problem occurs in an embedded database process, the process terminates abnormally. The embedded database then starts the recovery process and executes the rollback procedure. Because the process in question restarts automatically, the embedded database system administrator is not required to take any action. However, depending on the extent of the problem, the embedded database might terminate abnormally. If it does terminate abnormally, the embedded database restarts automatically. The embedded database system administrator must then eliminate the cause of the problem, and re-execute the operation.