Problem with creation of large heap- and javadump files

[JoOne]

Yesterday, I discovered several heapdump.[datestamp].######.####.####.phd files on our application servers and content manager server. Each file was approx. 900 MB in size and they were starting to fill up the disk.

I wonder if anyone know why these files are being created? There must be some problems with the java processes, causing these to dump the memory to a file. I guess that it has something to do with the content manager service and the reportnet dispatcher service (BIBusTkServerMain), but I´m not sure.

Our two application servers have 3,4 GB of RAM. In Cognos Configuration, these two servers have the "Medium configuration" set under "IBM Cognos 8 service". Should it be set to "Large configuration"? Can this be the issue here?

If anyone know something about this, I would greatly appreciate it.

Best regards,
Johan

[josepherwin]

One would not be able to determine what's causing the dump file to occur.

You will need to forward it to the IBM Cognos support for them to investigate.
I would assume that there is something wrong with your config files (not necessarily your Cognos Configuration)

However, you can disable the javadump by modifying cclWin32SEHConfig.xml under ../crn/configuration and set

nv_var name="CCL_HWE_ABORT" value="2:0"

OR

nv_var name="CCL_HWE_ABORT" value="0:0"

[JoOne]

Hello,

I have logged an SR to IBM Cognos, waiting for a reply from them.

Thanks for the tip for disabling the java dump, I will look it up.

[lynwell07]

Why that files are being created? There must be some problems with the java processes, causing these to dump the memory to a file. I guess that it has something to do with the content manager service and the reportnet dispatcher service?




_________________
Multivariate testing

[JoOne]

I think that the Maximum memory in MB value was set too low on our CM server. We lowered the value after some problems from 1152 to 768 (if I remember correct), but then we raised it again to 1024 and the problem with creation of core dump files seems to be resolved. Since then our CM server runs fine without any restarts.