Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] running top-level autogen.sh breaks romio in 1.6.3 tarball
From: Jeff Squyres (jsquyres_at_[hidden])
Date: 2012-11-01 09:11:41


Looks like this was already fixed in the trunk; I'm guessing there was a forgotten CMR to the v1.6 branch somewhere along the way.

I've now fixed it in the v1.6 branch; this fix will be in the nightly tarballs:

    https://svn.open-mpi.org/trac/ompi/changeset/27553

I'll CMR this to the v1.7 branch.

Since this type of error comes up periodically (i.e., someone has a legitimate reason to run autogen in a distribution tarball, only to find out that there's an autogen-required file missing from the tarball), I've added "run autogen" to Cisco's nightly Open MPI regression testing.

To anyone running MTT who also wants to test autogen on nightly tarballs, you can add

    ompi_autogen = 1

to your MTT OMPI "MPI Install" section.

On Oct 31, 2012, at 5:15 PM, David Shrader wrote:

> Hello,
>
> Thank you for the reply! All of the autotools I am using have the same or higher versions than those specified at http://www.open-mpi.org/software/ompi/v1.6/. I referenced the specific versions at the end of my initial email.
>
> After some digging on the svn branch and some help from Nathan Hjelm, I found and checked out ompi/mca/io/romio/romio/autogen.sh. If I put this file in the same location inside the tarball contents, romio is not broken by running the top-level autogen.sh. I am still using the same versions of autotools.
>
> Thank you for the reminder on not needing to run autogen.sh on the tarball version. Unfortunately, we're doing some modifications to the romio Makefile.am files to add another MPI-IO type and want to test against Open MPI releases. Hence, we have to regenerate all the configure scripts and Makefile.in files by running autogen.sh.
>
> Again, thank you all for your time!
> David
>
> On 10/31/2012 02:46 PM, Ralph Castain wrote:
>> We've seen this before - it's caused by using autotools that are too old. Please look at the HACKING file to see the required version levels.
>>
>> BTW: you should not be running autogen.sh on a tarball version. You should only run configure.
>>
>>
>> On Oct 31, 2012, at 1:31 PM, David Shrader <dshrader_at_[hidden]> wrote:
>>
>>> Hello,
>>>
>>> When using Open MPI from the 1.6.3 tarball, I have found that running the top-level autogen.sh breaks the romio component. Here are the steps to reproduce:
>>>
>>> 1) download openmpi-1.6.3.tar.bz2 from http://www.open-mpi.org/software/ompi/v1.6/
>>> 2) untar openmpi-1.6.3.tar.bz2
>>> 3) cd openmpi-1.6.3
>>> 4) ./autogen.sh
>>> 5) ./configure
>>>
>>> In configure's output, the following error can be seen:
>>>
>>> ...output snipped...
>>>
>>> *** Configuring ROMIO distribution
>>> configure: OMPI configuring in ompi/mca/io/romio/romio
>>> configure: running /bin/sh './configure' CFLAGS="-DNDEBUG -g -O2 -finline-functions -fno-strict-aliasing -pthread" CPPFLAGS=" -I/users/dshrader/tmp/openmpi-1.6.3/opal/mca/hwloc/hwloc132/hwloc/include -I/usr/include/infiniband -I/usr/include/infiniband" FFLAGS="" LDFLAGS=" " --enable-shared --disable-static --with-mpi=open_mpi --disable-aio --cache-file=/dev/null --srcdir=. --disable-option-checking
>>> Configuring with args CFLAGS=-DNDEBUG -g -O2 -finline-functions -fno-strict-aliasing -pthread CPPFLAGS= -I/users/dshrader/tmp/openmpi-1.6.3/opal/mca/hwloc/hwloc132/hwloc/include -I/usr/include/infiniband -I/usr/include/infiniband FFLAGS= LDFLAGS= --enable-shared --disable-static --with-mpi=open_mpi --disable-aio --cache-file=/dev/null --srcdir=. --disable-option-checking
>>> checking for Open MPI support files... in Open MPI source tree -- good
>>> ./configure: line 2805: PAC_PROG_MAKE: command not found
>>>
>>> ...output snipped...
>>>
>>> ./configure: line 7908: syntax error near unexpected token `newline'
>>> ./configure: line 7908: ` PAC_FUNC_NEEDS_DECL(#include <string.h>,strdup)'
>>> configure: /bin/sh './configure' *failed* for ompi/mca/io/romio/romio
>>> configure: WARNING: ROMIO distribution did not configure successfully
>>> checking if MCA component io:romio can compile... no
>>>
>>> ...remaining output snipped...
>>>
>>> None of the MPI/IO components work, including ufs, if I continue with a 'make' and 'make install'.
>>>
>>> Judging from the output from configure and a quick perusal of romio's configure script, it looks like some macros are not being correctly expanded in the creation of romio's configure script.
>>>
>>> For reference, here are the versions of my autotools:
>>> m4: 1.4.16
>>> autoconf: 2.69
>>> automake: 1.11.5
>>> libtool: 2.4.2
>>>
>>> I have not yet submitted this as a Trac item on svn.open-mpi.org. I wasn't sure what to put in the "Version" field as 1.6.3 wasn't listed there and I don't know if this is an issue in the 1.6 branch.
>>>
>>> Thank you very much for your time,
>>> David
>>>
>>> --
>>> David Shrader
>>> SICORP, Inc
>>> 1350 Central Ave
>>> Suite 104
>>> Los Alamos, NM 87544
>>> David.Shrader_at_[hidden]
>>>
>>> LANL contact information:
>>>
>>> LANL #: 505-664-0996
>>> LANL email: dshrader_at_[hidden]
>>>
>>> _______________________________________________
>>> 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
>
> --
> David Shrader
> SICORP, Inc
> 1350 Central Ave
> Suite 104
> Los Alamos, NM 87544
> David.Shrader_at_[hidden]
>
> LANL contact information:
>
> LANL #: 505-664-0996
> LANL email: dshrader_at_[hidden]
>
> _______________________________________________
> devel mailing list
> devel_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/devel

-- 
Jeff Squyres
jsquyres_at_[hidden]
For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/