Command terminated by signal 9

[paudirac]

I've been tryning to solve some system of diferential equations using
the gsl/odeiv library, but sometimes the program stops with

Command terminated by signal 9
Does anybody know what causes that?

Thanks,

[Måns Rullgård]

Post by paudirac
I've been tryning to solve some system of diferential equations using
the gsl/odeiv library, but sometimes the program stops with
Command terminated by signal 9
Does anybody know what causes that?

That's SIGKILL. Does your program use large amounts of memory? Which
operating system are you using?

[paudirac]

I'm using 2.4.20-20.9 GNU/Linux on a i686 and 2.6.8.1-12mdk GNU/Linux
on a i686 AMD Athlon(tm) 64 Processor 3400+.
I'm performing some operations on a an array of 400 doubles, but I
think this is not the cause of the problem, is it?
What other causes, apart of memory overflows, could cause such SIGKILL?
Thanks again,

[Måns Rullgård]

Post by paudirac
I'm using 2.4.20-20.9 GNU/Linux on a i686 and 2.6.8.1-12mdk GNU/Linux
on a i686 AMD Athlon(tm) 64 Processor 3400+.
I'm performing some operations on a an array of 400 doubles, but I
think this is not the cause of the problem, is it?
What other causes, apart of memory overflows, could cause such SIGKILL?

If Linux runs out of memory it will send a SIGKILL to the process(es)
using the most memory. There can probably be other reasons as well.
Check your system log for clues.

[Rich Teer]

Post by Måns Rullgård

Post by paudirac
I'm using 2.4.20-20.9 GNU/Linux on a i686 and 2.6.8.1-12mdk GNU/Linux
on a i686 AMD Athlon(tm) 64 Processor 3400+.
I'm performing some operations on a an array of 400 doubles, but I
think this is not the cause of the problem, is it?
What other causes, apart of memory overflows, could cause such SIGKILL?

If Linux runs out of memory it will send a SIGKILL to the process(es)

For this reason, it might be worth trying the same program
on Solaris x86 (recompiling it first, of course). The current
version of Solaris Express supports a 64-bit kernel on amd64
machines.

[Måns Rullgård]

Post by Rich Teer

Post by Måns Rullgård

Post by paudirac
I'm using 2.4.20-20.9 GNU/Linux on a i686 and 2.6.8.1-12mdk GNU/Linux
on a i686 AMD Athlon(tm) 64 Processor 3400+.
I'm performing some operations on a an array of 400 doubles, but I
think this is not the cause of the problem, is it?
What other causes, apart of memory overflows, could cause such SIGKILL?

If Linux runs out of memory it will send a SIGKILL to the process(es)

For this reason, it might be worth trying the same program
on Solaris x86 (recompiling it first, of course).

The quicker method of determining whether that was the cause is to
examine the output of "dmesg" after the program has been killed.
While it may be considered impolite of the Linux to kill processes in
that manner, it does make a note of what it did.

However, I somewhat doubt that memory shortage is the issue. 400
doubles is no large quantity.

[Dan Mercer]

"Måns Rullgård" <***@inprovide.com> wrote in message news:***@ford.inprovide.com...
: Rich Teer <***@rite-group.com> writes:
:
: > On Fri, 14 Jan 2005, Måns Rullgård wrote:
: >
: >> > I'm using 2.4.20-20.9 GNU/Linux on a i686 and 2.6.8.1-12mdk GNU/Linux
: >> > on a i686 AMD Athlon(tm) 64 Processor 3400+.
: >> > I'm performing some operations on a an array of 400 doubles, but I
: >> > think this is not the cause of the problem, is it?
: >> > What other causes, apart of memory overflows, could cause such SIGKILL?
: >>
: >> If Linux runs out of memory it will send a SIGKILL to the process(es)
: >
: > For this reason, it might be worth trying the same program
: > on Solaris x86 (recompiling it first, of course).
:
: The quicker method of determining whether that was the cause is to
: examine the output of "dmesg" after the program has been killed.
: While it may be considered impolite of the Linux to kill processes in
: that manner, it does make a note of what it did.
:
: However, I somewhat doubt that memory shortage is the issue. 400
: doubles is no large quantity.

Unless it's on the stack - are they automatic variables? More than once I
blew out the stack on HP-UX before I convinced the powers that be to
increase from the default settings.

Dan Mercer

:
: --
: Måns Rullgård
: ***@inprovide.com

[Artie Gold]

Post by Dan Mercer
: >
: >> > I'm using 2.4.20-20.9 GNU/Linux on a i686 and 2.6.8.1-12mdk GNU/Linux
: >> > on a i686 AMD Athlon(tm) 64 Processor 3400+.
: >> > I'm performing some operations on a an array of 400 doubles, but I
: >> > think this is not the cause of the problem, is it?
: >> > What other causes, apart of memory overflows, could cause such SIGKILL?
: >>
: >> If Linux runs out of memory it will send a SIGKILL to the process(es)
: >
: > For this reason, it might be worth trying the same program
: > on Solaris x86 (recompiling it first, of course).
: The quicker method of determining whether that was the cause is to
: examine the output of "dmesg" after the program has been killed.
: While it may be considered impolite of the Linux to kill processes in
: that manner, it does make a note of what it did.
: However, I somewhat doubt that memory shortage is the issue. 400
: doubles is no large quantity.
Unless it's on the stack - are they automatic variables? More than once I
blew out the stack on HP-UX before I convinced the powers that be to
increase from the default settings.

But it would be unlikely that theat would generate a SIGKILL -- a
SIGSEGV being much more likely.

HTH,
--ag

[Chuck Dillon]

Post by paudirac
I've been tryning to solve some system of diferential equations using
the gsl/odeiv library, but sometimes the program stops with
Command terminated by signal 9
Does anybody know what causes that?
Thanks,

Check the limit or ulimit -a values depending on you shell. If the
algorithm is recursive you could be hitting a limit. Increase any
suspect limits and see if it helps.

-- ced

[Heny Townsend]

Post by paudirac
I've been tryning to solve some system of diferential equations using
the gsl/odeiv library, but sometimes the program stops with
Command terminated by signal 9
Does anybody know what causes that?

ISTR this is being typically a spurious message generated by gdb. I.e.
it doesn't mean the process was terminated by signal 9, it means the
core dump was damaged and/or gdb had trouble reading it and ends up
printing the wrong error message. Are you using gdb when you get the
message? This is not from direct experience, it's something I remember
reading about. So google may help.

[Randy Howard]

Post by paudirac
I've been tryning to solve some system of diferential equations using
the gsl/odeiv library, but sometimes the program stops with
Command terminated by signal 9
Does anybody know what causes that?

# kill -9 <pid>