Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |  

This web mail archive is frozen.

This page is part of a frozen web archive of this mailing list.

You can still navigate around this archive, but know that no new mails have been added to it since July of 2016.

Click here to be taken to the new web archives of this list; it includes all the mails that are in this frozen archive plus all new mails that have been sent to the list since it was migrated to the new archives.

Subject: Re: [OMPI devel] openmpi with FT enabled
From: Ralph Castain (rhc_at_[hidden])
Date: 2013-10-28 13:51:55


Unfortunately, the entire infrastructure is currently broken. :-(

The values in that file are MCA parameters - the C code just checks to see which params are set in various places and takes appropriate action.

We planned to do a "reset" to re-enable C/R since ORTE is now completely async, the MPI layer is headed that way, and we need to add in the external checkpoint integration requested for the next-gen large-scale systems. If you'd like to help with that, you are certainly welcome! If so, we should probably chat off-list about it.

On Oct 28, 2013, at 10:22 AM, Adrian Reber <adrian_at_[hidden]> wrote:

> Thanks. That's okay. That's why I am looking at it. To get it working.
> It is just not clear how the values from ft-enable-cr are supposed to be
> used. They are used in the code as if they are variables. They are
> defined, however, in a text file (which is not in C). My question is
> how are the values from ft-enable-cr supposed to be used in the C code?
> Or is the code infrastructure used in the C/R part also completely broken?
>
> Adrian
>
> On Mon, Oct 28, 2013 at 09:39:19AM -0700, Ralph Castain wrote:
>> I'm afraid C/R isn't supported in the trunk nor 1.7 release series at this time. We're looking to restore that support next year as part of the 1.9 release series.
>>
>>
>> On Oct 28, 2013, at 8:47 AM, Adrian Reber <adrian_at_[hidden]> wrote:
>>
>>> I am trying to compile openmpi (Revision: 29539) from svn
>>> with '--with-ft=cr'. I get a compilation error and I am
>>> lost how to solve it:
>>>
>>> ../../../../opal/mca/base/mca_base_components_open.c: In function 'open_components':
>>> ../../../../opal/mca/base/mca_base_components_open.c:144:9: error: 'mca_base_component_distill_checkpoint_ready' undeclared (first use in this function)
>>> if (mca_base_component_distill_checkpoint_ready) {
>>> ^
>>> ../../../../opal/mca/base/mca_base_components_open.c:144:9: note: each undeclared identifier is reported only once for each function it appears in
>>>
>>> mca_base_component_distill_checkpoint_ready exists in
>>>
>>> contrib/amca-param-sets/ft-enable-cr
>>>
>>> but I am not sure if contents from this file can be used in the code or
>>> not. Any pointers on how to solve this error would be welcome.
>>>
>>> Adrian
>>> _______________________________________________
>>> devel mailing list
>>> devel_at_[hidden]
>>> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>>
>> _______________________________________________
>> devel mailing list
>> devel_at_[hidden]
>> http://www.open-mpi.org/mailman/listinfo.cgi/devel
> _______________________________________________
> devel mailing list
> devel_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/devel