Our system is on AIX5.3 ML5, DB2 v8.2 FP11, and we are running C applications.
There are 2 scheduled cronjobs, which are actually triggering the same binary, but passing in different parameter. With different parameters given, these 2 jobs are meant to do delete unused records from 2 different tables (there's no relationship between these 2 tables).
The scenerio is when the first job is still running (it takes longer when there are more records to be deleted), the second job get triggered. Looking at the application logging, it seems like the first job just stop there, and waiting for the second job to run finish, then it will continue running. But when it continues, it become very slow. Sometimes it managed to finish running, sometimes it'll exits with error.
There's no lock escalation in DB2 log files, but the number of lockheld keep increasing. The application is actually doing batch commit, but somehow the commit count doesn't increase while it keep taking up locks. No error captured in the db2diag.log either even when the application exits with error.
Anyone come across anything like this?
Also, wonder why the locktimeout didn't work as the application doesn't time out, but keep waiting until the second job finish, then continue to run again.