[mpich2-dev] Parent terminates when the spawned child terminates

[Lisandro Dalcin]

On 16 December 2010 12:36, Pavan Balaji <balaji at mcs.anl.gov> wrote:
>
> On 12/16/2010 06:26 AM, Suraj Prabhakaran wrote:
>>>>
>>>> If you see here, the child's first message will be displayed while the
>>>> second printf will not be displayed! Which shows that it cant get the
>>>> parent
>>>> anymore. And when a exit() happens abruptly, it shouldn't affect the
>>>> parent
>>>> rite (since they now dont have a common communicator, and without any
>>>> auto
>>>> clean option) ?
>>>> Please correct me if I am wrong.
>>>>
>>> I believe you are right.  For this program, at child exit() time
>>> parent and child are disconnected, and I understand the MPI standard
>>> says that the parent should not be affected. Of course, I'm not a MPI
>>> lawyer, and perhaps I misunderstood the wording...
>>
>> That implies that its a bug? I would benefit from a fix for this..
>
> No, AFAICT the MPI standard doesn't make such claim for spawned processes
> (unlike applications connected using connect/accept). If you want one
> process dying to not kill the remaining processes, you should use
> -disable-auto-cleanup.
>

Sorry, but I have to disagree. My reading of the first paragraph at
http://www.mpi-forum.org/docs/mpi22-report/node226.htm#Node226,
specifically the sentence below:

"""
Similarly, it might be desirable for a parent and child to disconnect,
so that errors in the child do not affect the parent, or vice-versa.
"""

means to me that the following comments also apply to spawned
processes. Am I missing something?



-- 
Lisandro Dalcin
---------------
CIMEC (INTEC/CONICET-UNL)
Predio CONICET-Santa Fe
Colectora RN 168 Km 472, Paraje El Pozo
Tel: +54-342-4511594 (ext 1011)
Tel/Fax: +54-342-4511169