Unanswered: instance crash after running a query on DB2 9.5 FP4
Does anyone have DB295 FixPack4? I am running the following query on it (XP SP2 or 3 local machine):
SELECT R.ROUTINESCHEMA, R.ROUTINENAME, R.SPECIFICNAME, R1.ROUTINESCHEMA, R1.ROUTINENAME, R1.SPECIFICNAME, R1.ROUTINETYPE FROM SYSCAT.ROUTINEDEP D, SYSCAT.ROUTINES R, SYSCAT.ROUTINES R1 WHERE D.ROUTINESCHEMA = R1.ROUTINESCHEMA AND D.BNAME = R.SPECIFICNAME AND D.ROUTINENAME = R1.SPECIFICNAME AND D.BTYPE IN ('F') AND R.ROUTINETYPE='P' AND R.ROUTINESCHEMA='DB2ADMIN' AND R.SPECIFICNAME='SQL090916170920800' AND R.ROUTINENAME='SP_PERFCNTR_SECONDARY' ORDER BY R.ROUTINESCHEMA, R.ROUTINENAME, R.SPECIFICNAME
where R.ROUTINESCHEMA='DB2ADMIN' is the schema of a procedure (any of them); R.SPECIFICNAME='SQL090916170920800' is the specific name of the procedure and R.ROUTINENAME='SP_PERFCNTR_SECONDARY' is the actual procedure name. Again, it does not matter what the procedure's body.
Must specify that the client version does not matter here.
The odd thing is that this query runs successfully on any other DB2 v9 version. (tried db295 FP3, etc.)
Can someone run this query as well for one of their procedures on DB2 v9.5 FP4? My server instance crashes, but gets back to normal when I restart it using "db2start".
PS: the error shown in server log is: "RETCODE : ZRC=0x87120007=-2028863481=SQLR_SEVERE_PGM_ERROR "Severe programming error" DIA8516C A severe internal processing error has occurred. " and it happens after this query.
The trap file should tell why the instance is crashing. In the diagpath, look for a dir that starts with FODC (you may see several of them - DB2 will create a new one for each crash occurrence). Inside the FODC dir, look for a trap file (*.trap.bin). Format it using db2xprt (for example: db2xprt *.trap.bin *.trap.fmt). Open the formatted output and look for "Failing instruction". This line will give you the function name where DB2 trapped. Then, use this function name to search for any known APARs here: IBM Software Support - DB2 for Linux UNIX and Windows - Overview
i don't see any bin file in FODC trap but lot of txt files and crash files which files need to be check to know why was the instance down
On Unix/Linux, trap files are readable and have .txt extension. You have several trap files on different partitions so you need to check which PID received a signal in the db2diag.log and then check its trap file. The most common signal is "Signal #11" (you can search trap files for this info), but it really depends on the problem. The best is to open a PMR to address the crash.