[yt-users] curious warning message in parallelHOP

John Wise jwise at physics.gatech.edu
Wed May 9 17:37:06 PDT 2012


Hi Matt and Geoffrey,

If you're talking to this John, I didn't make that change.  I think 
Britton made that change in changeset ab0e7664f75d.

But if popen() creates forks, I'd -1 on this being default, also because 
it causes problems in some MPI implementations, as Geoffrey has found 
out the hard way!

John

On 05/09/2012 08:23 PM, Matthew Turk wrote:
> Hi Geoffrey,
>
> Grepping through the yt codebase, I think the most likely culprit is
> that get_memory_usage (which I believe gets called often during
> parallel HOP) falls back on calling popen, which results in a fork.  I
> think John might have added this.  I am -1 on this being default
> behavior, and +1 on changing fallback (by default) to just return 0 if
> it can't get memory information from the resource module, and maybe
> governing this with a config parameter.  John, what do you think?
>
> -MAtt
>
> On Wed, May 9, 2012 at 7:46 PM, Geoffrey So<gsiisg at gmail.com>  wrote:
>> I got this warning message when running paralllel HOP, which I guess was
>> there before but never noticed before because it's close to the top (usually
>> looking at the bottom of the logs).  I was wondering if there's anything I
>> need to worry about?
>>
>> --------------------------------------------------------------------------
>> An MPI process has executed an operation involving a call to the
>> "fork()" system call to create a child process.  Open MPI is currently
>> operating in a condition that could result in memory corruption or
>> other system errors; your MPI job may hang, crash, or produce silent
>> data corruption.  The use of fork() (or system() or other calls that
>> create child processes) is strongly discouraged.
>>
>> The process that invoked fork was:
>>
>>    Local host:          lens51 (PID 6818)
>>    MPI_COMM_WORLD rank: 323
>>
>> If you are *absolutely sure* that your application will successfully
>> and correctly survive a call to fork(), you may disable this warning
>> by setting the mpi_warn_on_fork MCA parameter to 0.
>> --------------------------------------------------------------------------
>> [lens6:19012] 447 more processes have sent help message help-mpi-runtime.txt
>> / mpi_init:warn-fork
>> [lens6:19012] Set MCA parameter "orte_base_help_aggregate" to 0 to see all
>> help / error messages
>> [lens6:19012] 32 more processes have sent help message help-mpi-runtime.txt
>> / mpi_init:warn-fork
>> [lens6:19012] 32 more processes have sent help message help-mpi-runtime.txt
>> / mpi_init:warn-fork
>>
>> From
>> G.S.
>>
>> _______________________________________________
>> yt-users mailing list
>> yt-users at lists.spacepope.org
>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
>>
> _______________________________________________
> yt-users mailing list
> yt-users at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

-- 
John Wise
Assistant Professor of Physics
Center for Relativistic Astrophysics, Georgia Tech



More information about the yt-users mailing list