Few days back during Load Testing on performance environment I came across one issue. The issue was as follows:
When I connect to the application then initially it works fine. But in performance environment as the number of users connecting to the applications
increases at one point all the connections got terminated giving the following error:
com.s1.arch.persistence.exception.PersistenceException: The current transaction was rolled back because of error “-430”.. SQLCODE=-1476, SQLSTATE=40506, DRIVER=3.51.90
IBM InfoSphere Federation Server with IBM DB2 9.7 FP6
In this case the federated tables are not being able to access through nicknames.While searching out for this
I came across the following possible causes that might be responsible for this situation:
1. The Temporary Tablespace might be full.
2. The application Heap might got full because of which application was unable to process the sql statements.
But both these was not the cause since temporary tablespace was system managed and the APP_HEAP was AUTO-RESIZE.
Tablespace can be cross-verified using the syscat.tablespaces and APP_HEAP was checked in the database configuration parameters.
When I delved more into db2diag.log, I came to know about the actual cause behind this.
The issue was that the resources memory used by the processes was very low.
Due to limited resources db2fmp process was unable to create the new db2fmp thread which were causing the db2fmp unstable.
Because of which we were unable to login to the application after sometime.