F60WEBMX PROCESSES 의 UNIX CALL STACK 생성방법

Question

제품: AOL
작성날짜 : 2006-05-30
F60WEBMX PROCESSES 의 UNIX CALL STACK 생성방법
==========================================
PURPOSE
F60webmx process 가 CPU을 많이 차지하는 경우가 간혹 있으며
SQL trace외에 UNIX call stack으로 위 Process가 현재 Active 상태인지
혹은 Dump 등을 분석시 UNIX trace를 생성하는 방법니다.
Explanation
먼저 아래 명령어와 같이 Trace 를 실행합니다.
Log on as root.
# trace -a -L 4000000 -T 2000000 o ./trace.out;
# owsctl start admin
# trcstop
# trcrpt -O exec=on,pid=on trace.out > trcrpt.1
그다음 아래단계처럼 해당 Process를 찾습니다.
1) PS 명령어로 Process를 찾습니다.
          - ps -ef | grep -i f60webmx | grep webfile
     This will return an output like this:
          applmgr 6324 1052 0 14:34:56 ? 0:01 f60webmx webfile=5,1,gjimenez-sun     _14007_PROD
2) TUSC 를 실행합니다.
Execute your TUSC command for your process id identified on the previous step. In the above example, the PID is 6324.
3) 문제를 Reproduce하면서 Webfile 1번과 2번을 유의하여 Monitoring 합니다.
Ask any of your users in the location where you are having problems to log in to the application (Forms). Then,
make sure that a new process has been spawn (in the above example, a new f60webmx process should be spawn with a the parameter
webfile increased by 1: f60webmx webfile=5,2). This will tell us that the process #1 has been taken by the user we expect.
4) 문제가 재현될때까지 기다립니다.
Wait until the error reproduces for this user. Once the error surfaces, the TUSC process will die all by itself. Make sure you
upload the output file through Metalink for further examination.
Reference Documents
Note 244072.1

Answer 1

Always paste only the ERROR lines, not the full content.
Suggestions: Upgrade your Database for 9i to 10g (9i is out of support)
                      Size your hardware and application based on your system load etc
Solution for above issue: Try to kill some of the work processes from OS level (Application restart is better than this)
Regards,
Nick Loy