Re: /core. .... .dmp file

[Sam_L <lennon_s_j@xxxxxxxxxxx>]

I'm familiar with core dumps (used to program in mainframe 360 assembler 
when I was much younger).  What I don't like it the lack of control on 
where these dumps go and the fact that I don't know they occurred.

We had a writer that produces PDFs mysteriously end and it might be 
related.  For now I'm going to ignore this one and we should be applying 
cumes this month so maybe it won't happen again.

Sam

On 10/2/2012 9:38 AM, Scott Klement wrote:
> hi Sam,
>
> A core file occurs when a Unix program (PASE) crashes.  In order to
> provide diagnostic and/or debugging information for the problem, the
> system saves the program's raw memory to a file.  A debugging tool can
> open up that core file and can tell what statement the program was on,
> the value of all of the variables, what the error was, etc.
>
> In the original days of Unix, RAM technology used tiny little magnetic
> "cores" to store data.  Data that was in memory (vs. disk) would be said
> to be "in core".  That's why it's called a "core dump".  But it's really
> just raw memory written to a file.
>
> In many ways, it's similar to a "program dump" in the traditional
> environment.  Though, it's not a simple text document like a dump would be.
>
> Since recent Java JVMs are, indeed, PASE programs, it could've been Java
> that created this.  It could also be any other PASE program that you are
> running on the system.
>
> I don't know if there's any way to control where it goes?
>
>
> On 10/1/2012 4:48 PM, Sam_L wrote:
> > I ran into a 2GB IFS file today:
> >
> >     /core.20121001.094502.597368.0001.dmp.
> >
> > Anyone know where this might have come from?
> >
> > It was created at 9:45 this morning, but I see nothing in QHST that
> > seems to explain it.  I discovered it only because Mimix wasn’t happy
> > about it, otherwise I might never have known about it.  I definitely
> > don’t need it on the backup box.
> >
> > I imagine is it system generated--I’m guessing it is from Java and I've
> > found a similar /java, ... .dmp one, though much smaller.  We have
> > relatively little Java code, but I believe the PSF/400 PDFing routines
> > are in Java and we do a fair amount of that.
> >
> > If I must have these dumps, I really would like then to go somewhere
> > else so that I could better suppress replication.