Memory fault core dump

[Zuraina Sufar]

We're running on Tru64 Unix ver 5.1. We have this program that has been running in server A for quite a while. However, the program failed recently and still failing until now - it was terminated with Memory fault core dump. There was no system change/upgrade and it has been up for more than 20 days.
We copied the program to Server B and it works perfectly. Server A has more CPU/RAM as compared to server B. Both have similar apps environment.
What could have caused this error to happen?

[Joseph Loo]

hi,

not sure what applications u r running? however, to check the core dump:

# file core
what is the output for this?

if it is not inconvenient, what r the apps running?

regards.
(p.s. please assign points to those who have help u in your questions.)

[Bharat Katkar]

Another things would be to look for application generated log files if any. Also see if you can see anything out there in System Log files.

Hope that helps.
Regards,

[Manish Srivastava]

Hi,

The file output on the core file should tell what type of error occured. Tru64 is very lenient on apps and many a times SIGSEGV (segmentation violation) and bus errors go unnoticed. You can use gdb and get the stack trace if the app is not stiripped of symbols to locate the point where the failure occured.

manish

[Zuraina Sufar]

Thanks for the fast response. It is a reporting program written in ProC and compiled using Oracle ProC compiler, accessing Oracle DB (Oracle ver - 8.0.6.3)
(That's abt all I know abt the program - as I am only the Sys Admin).
Have not get the chance to get the output for command " #file core" as the file has been deleted. Will update on that.