On Nov 28, 2005, at 3:42 PM, Phil Yoon wrote:
> When I built lam v.7.1.1 on GNU/Linux, I used an additional
> configure flag, --with-memory-manager=none. So, I managed
> to build static binaries. However, when running a parallel
> job the following error message was issued. I'm wondering
> if the error is related to the disabled memory manager,
> or something else.
>
> -----------------------------------------------------------------------
> ------
> PID 27165 failed on node n0 (172.20.1.122) due to signal
> 11.
> -----------------------------------------------------------------------
> ------
This indicates that [at least] one of your processes seg faulted -- did
you get a core file? If so, can you get a back trace on it to see
where it died?
What RPI were you using? Note that using the ib or gm RPI *requires*
to use of the memory manager.
--
{+} Jeff Squyres
{+} The Open MPI Project
{+} http://www.open-mpi.org/
|
