Document Type | Troubleshooting
Category | Monitoring/Inspection
Applicable Product Versions | 6FS06, 6FS07, 6FS07PS, 7FS01, 7FS02, 7FS02PS
Document Number | TMOTS002
Issue
When attempting to collect a dump using cepa_control for analysis of the tbepa process in an AIX environment, the dump is not created and the tbepa process terminates (is killed).
$ ps -ef | grep 34931580 tibero 34931580 1 0 11:14:07 - 0:00 tbepa 48 1 180 $ cepa_control -l 5 -p 34931580 Log level change request was set. Sent a request to tbepa(34931580) $ ps -ef | grep 34931580 *tbepa process not found, killed after running cepa_control SQL> select ext_find_max(10,20) from dual ; TBR-2131: Generic I/O error. *The session connected to the tbepa process is terminated
Cause
When cepa_control is executed, it sends a sigusr1 signal to the tbepa process to change the log level or generate a memory dump. However, the cepa_control script is hardcoded to use kill -10 based on Linux, and in the AIX environment, kill -10 acts as a sigbus signal causing the process to terminate.
Solutions
Modify the hardcoded part to send SIGUSR1. (Applied patch: 326862)
CautionApply the patch through technical support provided by TmaxTibero.
NoteIn AIX or SunOS environments, modify the line 'kill -10 $EPA_PID' to 'kill -s SIGUSR1 $EPA_PID' in $TB_HOME/client/bin/cepa_control.$ vi $TB_HOME/client/bin/cepa_control ...omitted # Send a signal. Signo(10) is SIGUSR1 # kill -10 $EPA_PID kill -s SIGUSR1 $EPA_PID echo "Sent a request to tbepa($EPA_PID)" ...omitted