A UTM application is terminated abnormally by any of the following events:
internal UTM error
error in the system environment and shutdown of the Unix or Linux system
UTM administration command KDCSHUT KILL (or by the corresponding WinAdmin/WebAdmin or KDCADMI function)
user error
The following actions are performed when a UTM application is terminated abnormally:
All transactions currently being processed by the individual work processes are aborted immediately.
The connections to all communication partners of the application are shut down.
A UTM-specific dump is created for each work process of the application. See also the openUTM manual “Messages, Debugging and Diagnostics on Unix, Linux and Windows Systems”.
All processes of the application are terminated and all files are closed. No attempt is made to bring the KDCFILE to a consistent state. This does not occur until the application is restarted.
Following an abnormal termination of the application, you must first determine the cause of the crash. To find the cause, look for message K060 in the log of the work process on stdout. This message contains the dump error code as an insert. This error code gives precise information regarding the cause of the abnormal termination. You can also find the cause for the dump as part of the name of the UTM dump file. The meanings of the dump error codes are described in the openUTM manual “Messages, Debugging and Diagnostics on Unix, Linux and Windows Systems”. There are three possibilities:
The dump error code indicates that a KDCDEF operand must be modified.In this case, the KDCFILE must be regenerated. If you want to retain the application data in the page pool, proceed as follows:
warm start with ASYNTASKS=0, TASKS=1
terminate the application normally with KDCSHUT NORMAL
save the old KDCFILE
new KDCDEF generation with the modified operand
transfer the application data from the old to the new KDCFILE using KDCUPD
start the application with the new, updated KDCFILE
The dump error code cites the cause as:
a memory bottleneck
database is currently unavailable
When the error has been rectified, you can restart the application, and openUTM executes a warm start automatically.
A system error has occurred. In this case, produce diagnostic documentation and write a problem report to the system support personnel. To do this, you must edit the UTM dumps of all work processes of the application using the KDCDUMP tool. Further documentation includes the stdout and stderr system files, the gcores (for Unix and Linux systems), the utmwork program, the KDCDEF control statements, and an evaluation of the system log file.
A warm start with the same KDCFILE is not always successful in this case. If a warm start cannot be performed, you must regenerate the KDCFILE using KDCDEF.
If the application terminates abnormally, the KDCREM tool must be called before restarting the application (see following section).