[Info-ingres] Crash dumps

[Karl and Betty Schendel]

> On Aug 22, 2017, at 4:16 AM, Roy Hann <specially at processed.almost.meat> wrote:
> 
> I have an Ingres system running on RHEL and I have II_SEGV_COREDUMP=yes.
> I want to make sure that Ingres really will write a core file and
> that the OS is configured to process it. If it is, I want to find out
> where it puts the blessed thing.
> 
> In /proc/sys/kernel/core_pattern I see:
> 
>  |/usr/libexec/abrt-hook-ccpp %s %c %p %u %g %t e


Ugh.  That is some sort of red-hat-ish core dump interceptor.  We've run into it
a few times, but (unfortunately for you) always at a site where someone knew
how to turn it off without instructions from us.  It's pretty annoying.  I think there
might be some other variable specifying the max size of a core dump and you
might want to google for that one as well.

I don't run RH myself, so alas I can't tell you how to deal with it, and I'm quite
sure that your google skills equal or exceed mine.


> 
> Nothing seems to be written. 
> 
> I want to investigate using a development system but to test it 
> I need still to present a plan.
> 
> Will trace point SC908 crash the server in a way that generates a
> core dump? Or should I use kill -SIGSEGV? 

I don't believe SC908 does a core dump crash, although I'd have to try it
to be sure.  I suggest a kill -SEGV.

Karl