Confused...
Reg Clemens
reg at dwf.com
Wed Jul 12 19:00:28 UTC 2006
> Reg, by experiment, I can reproduce the 'Killed' error
> if I reduce the vmemoryuse limit. Have the default
> limits changed between updates?
>
> [root at dev1 ~]# cat hello.c
> #include <stdio.h>
> int i[100000000];
>
> main()
> {
> fprintf(stderr,"hello\n");
> while (0 == 0) {}
> }
> [root at dev1 ~]# limit
> cputime unlimited
> filesize unlimited
> datasize unlimited
> stacksize unlimited
> coredumpsize 0 kbytes
> memoryuse unlimited
> vmemoryuse 10240 kbytes
> descriptors 1024
> memorylocked 32 kbytes
> maxproc 12286
> [root at dev1 ~]# ./a.out
> Killed
>
> The coredumpsize limit of zero (default I think)
> would also prevent coredumps
>
> Cheers,
> Terry.
>
>
>
> >I have a program here that has been working for several years, and it suddenly
> >stopped working. Now it starts running and immediately shows 'Killed'.
> >
> >The very first line in the program is a fprintf(stderr, ...) and this message
> >does not get printed.
> >
> >And for reasons I dont understand, Linux is not dropping a core file.
> >
> >Ive recompiled, recompiled with -g, with -ggdb nothing.
> >
> >How do I get this thing to drop a core file?
> >Any possibility that one of the recent (last day or two) fc4 updates could
> >cause something like this?
> >
> >Confused.
> >--
Well, that looked like it could be it, the program is constantly generating
a larger dataset. The internal arrays are larger than needed but I thought
that perhaps as they were allocated I was getting problems.
But that doesnt seem to be it.
I set both vmemoryuse and stacksize to unlimited, and it still dies with
the Killed message. In fact, I took away its database, so that it was starting
over, and it also gets the Failed message.
It reads thru a FIFO, I deleted the FIFO and remade it, nada.
So still Confused.
--
Reg.Clemens
reg at dwf.com
More information about the fedora-list
mailing list