core dump

[txtraz]

Hi,
core dump. what is does mean & how do collect?

Please help me to collect.

Thanks;

[Gurudatta Shinde]

Hello,
The core dump file is collection of data when there is system crash due to lvm reason.Also it will keep the data which is lost.
If you had nickel script runnig on system,then just run that script and it will collect Core dump for you.

Regards,

The Guru.

[Jeeshan]

check this link


http://forums12.itrc.hp.com/service/forums/questionanswer.do?threadId=40484

[Venkatesh BL]

Check out manpage of 'core':

The HP-UX system writes out a file containing a core image of a
terminated process when certain signals are received (see signal(5)
for the list of reasons). The most common causes are memory
violations, illegal instructions, floating point exceptions, bus
errors, and user-generated quit signals. The core image file is
called core and is written in the process's working directory
(provided it is allowed by normal access controls). A process with an
effective user ID different from its real user ID does not produce a
core image.

[Steven E. Protter]

Shalom,

Core dumps vary a lot application to application.

what they are is a dump of whatever was in memory when the program unexpectedly terminates.

I've had reasonable results simply running the strings command against these things.

When Oracle core dumps, oracle support can analyze the dump and figure out if their code is bad or system setup needs to be changed.

So what to do and how useful it can be varies a lot, application to application.

Not being a developer, I assume this is functionality you have to write into the application, or it comes from the compiler.

SEP

[Dennis Handly]

>what is does mean & how do collect?

A core file is what you get if you don't have a signal handler for specific signals.
It captures the application's state at the time of the abort. With a debugger, you can use the core file to query the variables, get a stack trace, etc.

What do you mean by "collect"? You want to hunt them down and remove them? Or debug them?

>SEP: I've had reasonable results simply running the strings command against these things.

I would take that with a grain of salt. Looking at random memory will may give you all possible error messages, not the reason for the abort.
Using file(1) and gdb to get a stack trace would be most helpful.

>this is functionality you have to write into the application

No, this is part of the kernel. If you don't want a core file, you generally have to write extra code in the application.

[Dennis Handly]

Unless you are talking about a kernel crash or panic dump?
Those are in /var/adm/crash/.

[SKR_1]

You should contact various application support team to analys the core dumps.
For system core dumps you can contact HP, they will provide the support to you. For oracle dumps contact oracle.

Thanks

SKR