Dump files on the app server

[cognostechie]

I noticed that our app server is getting filled up by some 'dmp' files every day. These files are about 1 - 2 GB in size and it gets
created almost every day in the installation bin64 folder. This is the name of the file which got created today:


core.20140804.163744.3064.0001.dmp


Another text file gets created the same time as the dump file and this is the content of the 1st few lines:


0SECTION       TITLE subcomponent dump routine
NULL           ===============================
1TICHARSET     437
1TISIGINFO     Dump Event "corruptcache" (00400000) received
1TIDATETIME    Date:                 2014/08/04 at 16:37:57
1TIFILENAME    Javacore filename:    C:\Program Files\ibm\cognos\c10_64\bin64\javacore.20140804.163744.3064.0002.txt
1TIREQFLAGS    Request Flags: 0x81 (exclusive+preempt)
1TIPREPSTATE   Prep State: 0x106 (vm_access+exclusive_vm_access+trace_disabled)
NULL           ------------------------------------------------------------------------
0SECTION       GPINFO subcomponent dump routine
NULL           ================================
2XHOSLEVEL     OS Level         : Windows Server 2012 6.2 build 9200
2XHCPUS        Processors -
3XHCPUARCH       Architecture   : amd64
3XHNUMCPUS       How Many       : 8
3XHNUMASUP       NUMA is either not supported or has been disabled by user
NULL           
1XHERROR2      Register dump section only produced for SIGSEGV, SIGILL or SIGFPE.
NULL           
NULL           -------------------------------------------------------------------


Anyone seen this before? If so, What does it really mean?

[cognoslearner1]

You can delete those dmp files.
DMP files can be turned off by editing
configuration/cclWin32SEHConfig.xml
Change the value of CCL_HWE_ABORT to 0:0

[Suraj]

I've seen other dmp files but never encountered one that starts with core.

Usually, core dmp files are created when the server hangs or crashes.
Are you seeing any error in the system at around the time of that file creation?

If there was no one complaining, check logs around that time to see if there is any weird thing going on.

[cognostechie]

The log does not say anything for that time but the dump file says corrupt cache ! One thing we have which is wrong is that around 950 reports are scheduled to run at the same time which are fired up by a trigger !

The earlier BI team did not know how to resolve performance issues so they told the business to schedule the reports and save the output so that when the business goes into Cognos they would just click on the reports and get the saved outputs instead of having to watch the hourglass :)