How to analyze core dump generated by user application

[arunkrish]

My user application developed for ESXi 5.5 occasionally generates SIGSEGV signal which in turn causes core dump generated.

I have used this command to enable core dump generation in /var/core directory.

/sbin/vsish -e set /userworld/global/coreDumpEnabled 1

After having dump generated, I have tried using esxcfg-dumppart to extract log info, as per this link http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=100679...

But it seems that this tool is meant only for analyzing kernel dump. 

/var/core # esxcfg-dumppart -L appdump-zdump.000

Error running command. Unable to extract log. Couldn't find a valid VMKernel dump file. Dump partition might be uninitialized.

Is there any way to analyze core dump generated by userland application in ESXi 5.

Thanks,

Arun

[grasshopper · Accepted solution]

Hi arunkrish,

Correct, that tool (esxcfg-dumppart) is exclusively for extracting a readable text file from vmkernel-zdumps.  The preferred method for reviewing Guest OS crash dumps is vmss2core which gives you the desired crashdump info in a format readable by the preferred debugger for the guest OS in question.  As I understand it, the *zdump.000, *zdump.001, etc. are trapping issues at the VM level (i.e. as opposed to the GOS).

I think only VMware support can read those gos/vmx based dumps that you are currently trying to review (hopefully I'm wrong).  However, you can try your luck opening them in a pro debugger such as IDA Pro (speculation).  They have a free trial.